Full text of "A study of American intelligence"

See other formats

A STUDY OF AMERICAN
INTELLIGENCE

STUDY OF AMERICAN

INTELLIGENCE

By carl c! BRIGHAM, Ph.D.

ASSISTANT PROFESSOR OF PSYCHOLOGY
IN PRINCETON UNIVERSITY

A FOREWORD
By Robert M. Yerkes, Ph.D.

CHAIRMAN RESEARCH INFORMATION
service: NATIONAL RESEARCH COUNCIL

PRINCETON

Princeton University Press

LONDON: HUMPHREY MILFORD
OXFORD UNIVERSITY PRESS

1923

Second Printing

PRINTED AT THE PRINCETON UNIVERSITY PRESS, PRINCETON, U.S.A.

FOREWORD

Two extraordinarily important tasks confront our nation :
the protection and improvement of the moral, mental and
physical quality of its people and the re-shaping of its in-
dustrial system so that it shall promote justice and encour-
age creative and productive workmanship. I have been
asked to write this Foreword because of my official con-
nection, as chief of the Division of Psychology, Office of
the Surgeon General of the Army, with psychological ex-
amining during the war, but I have consented to write it
because of my intense interest in the practical problems
of immigration and my conviction that the psychological
data obtained in the army have important bearing on some
of them.

When in April, 1917, 1 visited Canada to learn what use
our neighbors were making of psychological principles and
methods in their military activities, I found Mr. Carl C.
Brigham attached as psychologist to the Military Hospitals
Commission. With him as my guide, I spent several hours
in interviewing military and civil officers and in discussing
our mutual problems and needs. The valuable information
which Mr. Brigham helped me to secure and his advice
contributed substantially to the report which I later pre-
sented to my professional colleagues at home, and to rep-
resentatives of the United States army.

vi AMERICAN INTELLIGENCE

In October, 1917, our friend, eager for larger opportuni-
ties for professional service than the Canadian army prom-
ised, accepted appointment in the Sanitary Corps of the
United States army for psychological service. He aided
efficiently in the trials of methods of examining at Camp
Dix, New Jersey, and he was then ordered to the office of
the Surgeon. General in Washington to help with the revis-
ion of tests and the preparation of new methods. Thus he
became thoroughly familiar with the procedures and results
of psychological examining in the army, while at the same
time contributing generously of ideas, labor and enthu-
siasm. With deep satisfaction I use this opportunity to men-
tion Mr. Brigham's national service and his exceptional
fitness to study and to discuss the relations of army meas-
urements of intelligence to nativity and residence.

It appears that Mr. Charles W. Gould, a clear, vigorous,
fearless thinker on problems of race characteristics, amal-
gamation of peoples and immigration, raised perplexing
questions which drove Mr. Brigham to his careful and
critical re-examination, analysis, and discussion of army
data concerning the relations of intelligence to nativity
and length of residence in the United States. In a recently
published book, America, A Family Matter, to which this
little book is a companion volume, Mr. Gould has pointed
the lessons of history for our nation and has argued strongly
for pure-bred races.

For the observational data which Mr. Brigham used in
preparing this book we are indebted to the competent and
devoted company of psychologists which during the war

FOREWORD vii

labored in camp and laboratory on the preparation of meth-
ods, the conduct of examinations, and the application of re-
sults. But the fruits of the labors of these many psychol-
ogists might have been lost to the world had it not been for
the insight, zeal, and industry of Carl R. Brown, Mark A.
May and Edwin G. Boring, who evolved methods of statis-
tical treatment, applied them and prepared the resulting
materials for publication.

Mr. Brigham has rendered a notable service to psychol-
ogy, to sociology, and above all to our law-makers by
carefully re-examining and re-presenting with illuminating
discussion the data relative to intelligence and nativity
first published in the official report of psychological exam-
ining in the United States army. Far from belittling or
casting doubt on the general reliability of the results con-
tained in the report, he has essentially confirmed the major
findings in the field of his special inquiry and has adduced
new evidences of the trustworthiness and scientific value
of the statistical methods used by military psychologists.
His task has been arduous and difficult, involving an im-
mense amount of tedious labor for mathematical calcula-
tions and critical study of results. The volume which is the
outcome of Mr. Brigham's inquiry, and which I now have
the responsibility and satisfaction of recommending, is sub-
stantial as to fact and important in its practical implica-
tions. It is not light or easy reading but it is better worth
re-reading and reflective pondering than any explicit dis-
cussion of immigration which I happen to know. The
author presents not theories or opinions but facts. It be-

viii AMERICAN INTELLIGENCE

hooves us to consider their rehabihty and their meaning,
for no one of us as a citizen can afford to ignore the menace
of race deterioration or the evident relations of immigra-
tion to national progress and welfare.

Robert M. Yerkes

Washington, D. C.
June 1922

CONTENTS

PAGE

Foreword v

Introduction xix

Part I : The Army Tests

Section 1 examination alpha 3

Section 2 examination beta 32

Section 3 the individual examinations 54

Section 4 reliability of the measures 59

Part II : Statistical Analysis of the Army Test
Results
Section 1 the principal sample 75

Section 2 analysis of the main groups of the

PRINCIPAL sample 77

Section 3 analysis of the white draft into

FOREIGN AND NATIVE BORN 84

Section 4 analysis of the foreign born white

DRAFT into YEARS OF RESIDENCE
GROUPS 88

Section 5 analysis of immigration to the

UNITED states 112

Section 6 analysis of the foreign born white

DRAFT BY COUNTRY OF BIRTH 118

Section 7 reliability of the results 154

Section 8 the race hypothesis 157

Section 9 re-examination of previous con-
clusions IN THE LIGHT OF THE RACE
HYPOTHESIS 177

Section 10 comparison of our results with the
conclusions of other writers on
the subject 182

Conclusions 197

PLATES

PLATE PAGE

I Alpha test 1 : oral directions 5

II Alpha test 2 : arithmetical reasoning 9

III Alpha test 3 : practical judgment 13

IV Alpha test 4 : synonym-antonym 18
V Alpha test 5 : disarranged sentences 21

VI Alpha test 6 : number series completion 24

VII Alpha test 7 : analogies 26

VIII Alpha test 8 : information 29

IX Beta test 1 : maze 35

X Beta test 2 : cube analysis 38

XI Beta test 3 : x-o series 41

XII Beta test 4 : digit-symbol 44

XIII Beta test 5 : number checking 47

XIV Beta test 6 : picture completion 50
XV Beta test 7 : geometrical construction 53

FIGURES

FIGURE PAGE

1. Distribution of scores of the oral directions test 6

2. The Gaussian normal distribution 7

3. Distribution of scores of the arithmetical reason-

ing test 10

4. Distribution of scores of the practical judgment

test 14

5. Distribution of scores of the synonym-antonym

test 17

6. Distribution of scores of the disarranged sentence

test 20

7. Distribution of scores of the number series com-

pletion test 23

8. Distribution of scores of the analogies test 25

9. Distribution of scores of the information test 28

10. Black-board chart for demonstrating the maze

test 33

11. Distribution of scores of the maze test 34

12. Black-board chart for demonstrating the cube

analysis test 36

13. Distribution of scores of the cube analysis test 37

14. Black-board chart for demonstrating the X-0

series test 39

15. Distribution of scores of the X-0 series test 40

16. Black-board chart for demonstrating the digit-

symbol test 42

xii FIGURES

FIGURE PAGE

17. Distribution of scores of the digit-symbol test 43

18. Black-board chart for demonstrating the number

checking test 45

19. Distribution of scores of the number checking test 46

20. Black-board chart for demonstrating the picture

completion test 48

21. Distribution of scores of the picture completion

test 49

22. Black-board chart for demonstrating the geomet-

rical construction test 51

23. Distribution of scores of the geometrical con-

struction test 52

24. The normal distribution curve 59

25. A skewed distribution curve 60

26. Examination alpha as independent of education 65

27. Distribution of intelligence scores according to

rank 66

28. Success in Officers' Training Camps as predicted

by examination alpha 67

29. Comparison of army test records with various

independent criteria 69

30. Success in civil occupations compared with army

test records 70

31. Distributions of scores of the white officers, white

draft, and negro draft on the combined scale 81

32. Distributions of scores of the native born and for-

eign born white draft on the combined scale 87

33. Apparently increasing average intelligence with

increasing years of residence 94

34. Distributions of the alpha scores of three groups 108

35. Analysis of immigration by countries 114

36. Relative standing of the nativity groups accord-

ing to their average intelHgence 124

FIGURES xiii

FIGURE PAGE

37. Relative standing of the nativity groups in the

proportions of A and B men, and D, D and

E men 146

38. The proportion of each nativity group obtaining

scores at or above the average of the white
ojQficers 149

39. The proportion of each nativity group at or below

the average of the negro draft 151

40. The proportion of each nativity group testing be-

low the approximate "mental age" of eight 153

41. Analysis of immigration to the United States ac-

cording to the estimated amount of Nordic,
Mediterranean and Alpine blood 164

42. Volume of immigration by decades 166

43. The distributions of the intelligence scores of the

Nordic, Mediterranean and Alpine groups 170

44. The distributions of the intelligence scores of the

English speaking Nordic and the non-English
speaking Nordic groups 173

45. The distributions of the intelligence scores of the

non-English speaking Nordic group and the
combined Mediterranean and Alpine groups 175

46. The decline of intelligence with each succeeding

period of immigration 198

47. The constituent elements of American intelhgence 200

TABLES

PAGE

1. Distribution of the intelligence scores of the main

groups of the principal sample on the combined
scale 80

2. Analysis of the white draft into foreign born and

native born groups 86

3. Analysis of the foreign born white draft by years

of residence in the United States 90

4. Comparison of the average scores on the combined

scale of the five years of residence groups of the
foreign born white draft 91

5. Comparison of the average scores on the combined

scale of the native born white draft with the five
years of residence groups of the foreign born
white draft 92

6 . Per cen t . that emigration was of immigration for 1 5

countries since 1908 98-99

7. Distribution of alpha scores of five groups 106

8. Per cent, of total immigration coming from var-

ious countries during periods roughly correspond-
ing to the five years of residence groups 113

9. Analysis of the foreign born white draft by coun-

try of birth : actual distributions 120-121

10. Analysis of the foreign born white draft by coun-

try of birth: percentage distributions 122-123

11. Differences between England and other countries 126

12. Differences between Scotland and other countries 127

13. Differences between Holland and other countries 128

14. Differences between Germany and other countries 129

15. Differences between the United States and other

countries 130

16. Differences between Denmark and other countries 131

17. Differences between Canada and other countries 132

xiv

TABLES XV

PAGE

18. Differences between Sweden and other countries 133

19. Differences between Norway and other countries 134

20. Differences between Belgium and other countries 135

21. Differences between Ireland and other countries 136

22. Differences between Austria and other countries 137

23. Differences between Turkey and other countries 138

24. Differences between Greece and other countries 139

25. Differences between Russia and other countries 140

26. Differences between Italy and other countries 141

27. Differences between Poland and other countries 142

28. Per cent, of each nativity group in the A and B

groups 144

29. Per cent, of each nativity group in the D, D

and E groups 145

30. Per cent, of each nativity group at or above the

average of the white officers 148

31. Per cent, of each nativity group at or below the

average of the negro draft 150

32. Per cent, of each nativity group below the approxi-

mate "mental age" of eight 152

33. Tentative estimates of the proportion of Nordic,

Alpine and Mediterranean blood in each of the
European countries 159

34. Arrivals of alien passengers and immigrants, 1820

to 1920 160-161

35. Estimate of the amount of Nordic, Mediterranean

and Alpine blood coming to this country from
Europe in each decade since 1840 163

36. Analysis of the foreign born white draft by races 169

37. Analysis of the Nordic sample into an English

speaking Nordic group and a non-English
speaking Nordic group 172

38. Population of the United States in 1920 203

ACKNOWLEDGMENTS

This study is a continuation of the work of the small
group of psychologists who carried on the difficult task of
analyzing the data from the army psychological examina-
tions in the office of the Surgeon General of the Army. My
presentation contains nothing new in methodology and is
merely an extension of lines of investigation suggested by
this group of workers. It rests on the foundations which
they built.

I wish to make especial acknowledgment to Colonel
Robert M. Yerkes, who has read the manuscript several
times in its various stages of preparation and has given
many helpful suggestions. Professor Carl R. Brown of the
University of Michigan, formerly of the Surgeon General's
staff, assisted me when I first began to use the combined
scale, and subsequently read Sections 1 to 7 of Part II in
manuscript. Professor E. G. Boring of Harvard University
read Sections 1 to 7 of Part II, and gave me invaluable
assistance, especially in the treatment of Section 4. Pro-
fessor Mark A. May of Syracuse University, Professors
Edwin G. Conklin and Howard C. Warren of Princeton
University read Sections 1 to 7 of Part II, and suggested
many important changes. Without the assistance of all of
these gentlemen I could not have carried through the task.

Mr. Charles W. Gould suggested this continuation of
the army investigations in the first instance, has sponsored
the work throughout, has read and re-read all of the manu-
script at every stage of its preparation, and is mainly
responsible for the whole work. In my treatment of the
race hypothesis I have relied on his judgment and on two
books, Mr. Madison Grant's Passing of the Great Race, and

xvii

xviii AMERICAN INTELLIGENCE

Professor William Z. Ripley's Races of Europe. These
three gentlemen cannot be held responsible however for my
percentage analysis of the present racial constitution of
European countries, an analysis which I made as a novice
in the field of anthropology, and for which I offer further
apologies in the text of Section 8.

Mr. David M. Maynard and Mr. Charles H. HeUiwell,
two of my ujidergraduate students, assisted me faithfully
in carrying through the laborious statistical calculations
involved in using the combined scale.

Carl C. Brigham

Princeton, N. J.
September 1922

INTRODUCTION

The question of the differences that may exist between
the various races of man, or between various sub-species
of the same race, or between pohtical aggregations of men
in nationahty groups may easily become the subject of
the most acrimonious discussion. The anthropologists of
France and Germany, shortly after the close of the Franco-
Prussian war, fought another national war on a small
scale. It is difficult to keep racial hatreds and antipathies
out of the most scholarly investigations in this field. The
debate becomes especially bitter when mental traits are
discussed. No one can become very indignant on finding
his race classified by its skull dimensions, stature, or hair
color, but let a person discover the statement that his race
is unintelligent or emotionally unstable, and he is immedi-
ately ready to do battle.

Until recent years we have had no methods available
for measuring mental traits scientifically, so that the lit-
erature on race differences consists largely of opinions of
students who are very apt to become biased, when, leaving
the solid realm of physical measurements, they enter the
more intangible field of estimating mental capacity.

Gradually, however, various investigators using more or
less refined psychological measurements commenced to as-
semble a body of data that will some day reach respectable
proportions. The status of the psychological investigations
of race differences up to 1910 has been admirably sum-
marized by Woodworth.i Since 1910, we have witnessed

IR. S. Woodworth. Racial Differences in Mental Traiis, Science, New Series, Vol.
31, pp. 171-186.

XIX

XX AMERICAN INTELLIGENCE

in this country a remarkable development in methods of
intelligence testing, and these methods have been applied
to the study of race differences. Scattered investigations
report and compare the intelligence scores of children of
white, negro, or Indian parentage, and sometimes the
scores of various nationality or nativity groups. The re-
sults of these investigations are, however, almost impos-
sible to correlate, for they have been made by different
methods, by different measuring scales, on children of a
wide variety of chronological ages, and above all, on com-
paratively small groups of subjects, so that conclusions
based on the studies have no high degree of reliability.

For our purposes in this country, the army mental tests
give us an opportunity for a national inventory of our own
mental capacity, and the mental capacity of those we have
invited to live with us. We find reported in Memoir XV of
the National Academy of Sciences i the intelligence scores
of about 81,000 native born Americans, 12,000 foreign
born individuals, and 23,000 negroes. From the standpoint
of the numbers examined, we have here an investigation
which, of course, surpasses in rehability all preceding in-
vestigations, assembled and correlated, a hundred fold.
These army data constitute the first really significant con-
tribution to the study of race differences in mental traits.
They give us a scientific basis for our conclusions.

When we consider the history of man during the half
million years which have probably elapsed since the time
of the erect primate, Pithecanthropus, the temporary polit-
ical organizations, such as Greece, Rome, and our modern
national groups, become of minor importance compared
with the movements of races and peoples that have oc-
curred. The tremendous expansion of the Alpine race at
the end of the Neolithic and the beginning of the Bronze

^^Psychological Examining in the United States Army. Edited by Robert M.
Yerkes. Washington: Government Printing Office, 1921, Pp. 890.

INTRODUCTION xxi

Period, the submergence of this race by the Nordics in the
2000 years preceding the Christian era, and the subsequent
peaceful re-conquest of Eastern Europe by the Alpine Slavs
from the Dark Ages on, represent an historical movement
in comparison with which the Great World War of 1914
resembles a petty family squabble.

If the history of the United States could be written in
terms of the movements of European peoples to this con-
tinent, the first stage represents a Nordic immigration, for
New England in Colonial times was populated by an almost
pure Nordic type. There followed then a period of Nordic
expansion. The next great movement consisted of the mi-
grations of Western European Mediterraneans and Alpines
from Ireland and Germany, a movement which started
about 1840, and which had practically stopped by 1890.
Since there is a considerable proportion of Nordic blood in
Ireland and Germany, we should not regard the original
Nordic immigration as a movement which stopped sud-
denly, but merely as having dwindled to two-fifths or one-
half of the total racial stock coming here between 1840 and
1890. The third and last great movement consisted of mi-
grations of the Alpine Slav and the Southern European
Mediterraneans to this continent, a movement that started
about 1890, and which has not yet ceased. Running parallel
with the movements of these European peoples, we have
the most sinister development in the history of this con-
tinent, the importation of the negro.

The army mental tests enable us to analyze the elements
entering into American intelligence. The intelligence test
records of the native born, the foreign born, and the negro
are at our disposal. The records deserve the most serious
study. But before considering the results of the army tests,
a person should be well informed concerning the nature of
the tests, and the manner in which they were constructed.

xxii AMERICAN INTELLIGENCE

The army psychological tests included three types of
examination :

(1) Group examination alpha, which included eight dif-
ferent sorts of tests, most of which involved the ability to
read English.

(2) Group examination beta, which included seven dif-
ferent sorts of tests, none of which involved the ability
either to read English or to imderstand spoken Enghsh,
the tests consisting of pictures, designs, etc., and being
given by instructions in pantomime.

(3) Individual examinations of two types:

(a) Those involving the use of English, the Stanford
revision of the Binet-Simon scale and the point
scale, and

(6) Those involving no English, consisting of con-
struction puzzles, etc., the instructions being
given by gestures, the "performance scale."

When a detachment reported for psychological examina-
tion, the first step was that of separating the English
speaking and literate from the non-English speaking or
iUiterate. Those who were both English speaking and liter-
ate were given examination alpha. All others were sent to
beta. At the close of examination alpha, all men who had
made low scores were sent to beta. After examination beta
had been given, the examiners tried to recall for individual
examinations all men who had made a low score in beta.
In the rush of examining it was impossible to recall all men
for individual examinations who should have been given
special examinations, and some men were graded on alpha
who should have been graded on beta, and vice versa, but

INTRODUCTION xxiii

most men were properly graded by the rough methods in
use. In each one of the examinations the range of scores
was so great that most men had an opportunity to score.

The great contribution of the committee that first de-
vised the army examining methods and of the men who
subsequently developed additional methods in the army
consisted of creating and standardizing group examinations
alpha and beta. The methods of individual examining were
already in existence, the Stanford-Binet scale being an elab-
oration of Binet's "mental age" scale, and the tests of the
performance scale having been more or less completely
worked out by other investigators. The task of examining
men in large groups was first carried through successfully
in the army. Before the war, many psychologists would
have scoffed at the notion of examining two or three hun-
dred men at once by giving them booklets containing dif-
ferent sorts of tests, but the large group examinations be-
came matters of daily routine. Group tests have subsequent-
ly been tried out in schools and industries with excellent
results from the standpoint of test administration. Indeed,
when the army alpha examination was given at Ohio State
University in October, 1919, practically the entire student
body, 6000 in number, was tested by five examiners in
eight hours. In the service, it was found that one examiner
could control a group of 200 men with ease. The alpha in-
structions were read by the examiner, and the men ordered
to start and stop at the proper time. Examination beta was
more difficult to administer, and was given to smaller
groups.

The statistical methods of treating the results of the
army tests used in this study are rather intricate, but the
principles involved are easily understood. At the outset we
must frankly admit that there were minor errors in the
three types of examinations given. We can not correct the

xxiv AMERICAN INTELLIGENCE

type of tests that were used, but we can correct the method
of scoring them. Most of the difficulties of scoring arise
from the fact that different types of measuring scales were
used. During the war, the different scales were converted
into one general scale of letter grades (A, B, C+, C, C ,
D and D ). This method was rough, and although it an-
swered the purposes of the army at the time, it can not be
used in any scientific interpretation of the results.

Examination alpha was scored by finding the score on
each of the eight tests, adding to get a total, and then con-
verting the total into a letter grade. Beta was similarly
scored. It is apparent that some tests in alpha might be
more difficult than others, that some tests in beta might be
easier than any test in alpha, and that variations might
have occurred which it was impossible to predict at the
time the examinations were made. Recognizing these facts,
then, the army statisticians worked out another method of
scoring the results, which eliminates all of these sources
of error. This method is known as the combined scale, a
theoretical intelligence scale running from to 25, into
which the alpha, beta and individual examination scores
may be converted, so that we finally have one measure-
ment instead of three.

Psychological measurements involve much more than
creating tests and giving tests. After all the results are in,
we still have the problem of interpreting the results, and
this interpretation is largely a statistical problem. Too
much credit can not be given to the staff of the Psycholog-
ical Division of the Surgeon General's Office, who con-
tinued in the service long after the war was over, patiently
studying and analyzing the results. The combined scale
was very largely the work of two young psychologists, Carl
R. Brown and Mark A. May, and their work on this prob-
lem, reported in Chapter 2, Part 3 of Memoir XV, is with-

INTRODUCTION xxv

out doubt the greatest contribution that has yet been made
to the statistical phases of the science of mental measure-
ment.

The theory underlying the combined scale is simply that
of regarding each test of alpha and beta as a separate
measuring scale. One group of individuals including 1047
men born in English speaking countries, was examined on
alpha, re-examined on beta, and if possible, examined again
on the Stanford-Binet scale. This group of 1047 cases con-
stituted the basis on which a method of combining the sep-
arate tests into a combined scale was empirically evolved.

From now on in the course of our study of the army test
records, we must regard alpha and beta as two booklets
containing, in all, fifteen different measuring scales of in-
telligence. The first step in the study consists of under-
standing the nature of each of the fifteen scales. In Part I,
the fifteen tests have been reproduced (Plates I to XV),
and the actual records of the 1047 men shown in each in-
stance, so that the reader may see exactly how the tests
worked.

PARTI
THE ARMY TESTS

SECTION I

EXAMINATION ALPHA

Alpha Test 1. Oral Directions

The first test in alpha consisted of a series of commands
or directions which were to be executed quickly. The in-
structions, with the incidental commands about stopping
and starting eliminated, are reproduced below. One may
read the instructions for each item to himself slowly and
turn the page to Plate I to test his own ability to execute
the commands.

Instructions : Oral Directions (Form 8)

Item 1. Time limit : 5 seconds.

"Make a figure 2 in the second circle and also a
cross in the third circle."

Item 2. Time limit : 5 seconds.

"Draw a line from circle 1 to circle 4 that will
pass below circle 2 and above circle 3."

Item 3. Time limit : 10 seconds.

"Make a figure 1 in the space which is in the
square but not in the triangle, and also make a
cross in the space which is in the triangle and in
the square."

Item 4. Time limit : 10 seconds.

"Make a figure 2 in the space which is in the
circle but not in the triangle or square, and also
make a figure 3 in the space which is in the tri-
angle and circle, but not in the square."

Item 5. Time limit : 10 seconds.

"If taps sound in the evening, then put a cross

4 AMERICAN INTELLIGENCE

in the first circle; if not, draw a line under the
word NO."

Item 6. Time limit : 10 seconds.

"Put in the first circle the right answer to the
question: *How many months has a year.?' In
the second circle do nothing, but in the fifth cir-
cle put any number that is a wrong answer to
the question that you have just answered cor-
rectly."

Item 7. Time limit : 10 seconds.

''Cross out the letter just after F arid also draw
a line under the second letter after I."

Item 8. Time limit : 10 seconds.

"Make in the first circle the last letter of the^r^^
word; in the second circle the middle letter of the
second word and in the third circle the first letter
of the third word."

Item 9. Time limit : 15 seconds.

''Cross out each number that is more than 50 but
less than 60."

Item 10. Time limit : 15 seconds.

"Put a 4 or a 5 in each of the two largest parts
and any number between 6 and 9 in the part next
in size to the smallest part."

Item 11. Time limit : 25 seconds.

"Draw a line through every odd number that is
not in a square, and also through every odd
number that is in a square with a letter."

Item 12. Time limit : 10 seconds.

"If 4 is more than 2, then cross out the number 3
unless 3 is more than 5, in which case draw a
line under the number 4."

1.O00OO

2 t^i@li)0

5 O O O ^'es /Vo

e OOOOO

7 ABGDEFGHIJKLMNOP

8 O O O MILITARY GUN CAMP

9 34-79-56.87-68-25-82-47.27-31-64.93-71-41-52-99

11 HAOA [I]A[I1

12 123456789

PZa^e I. Alpha Test 1 : Oral Directions.

6 AMERICAN INTELLIGENCE

In scoring the papers one point was given for each cor-
rect response. The group of 1047 individuals born in Eng-
Hsh speaking countries obtained the following scores:

Total score of test 12 3 4 5 6 78 9 10 11 12

Number who made each score ... 73 78 93 116 100 121 131 94 82 67 52 28 12

These scores are shown graphically in Figure 1, the
horizontal direction indicating the total score from the
lowest possible (0) to the highest possible (12), while

TEST 1,
ALPHA

100-

Figure 1. Distribution of scores of the Oral Directions test. (From
p.624<,MemoirXV.)

the vertical scale represents the number of cases getting
each score, 72 at 0, 78 at 1, etc.

For our purposes, we do not want a test that everyone
can pass, for if everyone passed, no one would be graded.
An ideal test would be one in which practically everyone
could obtain some score and which very few could finish.

AMERICAN INTELLIGENCE 7

Then all people would be measured. An ideal test would
also show a distribution of responses grouped symmetric-
ally about the average, for, as a general rule, all measures
of individual differences in mental traits show a distribution
similar to the normal probability or chance distribution.
An ideal test would give the type of distribution shown in
Figure 2.

200"

150-

lOG-

J L

50-

0-
(

n M .1 ...... n

Figure 2. The Gaussian normal distribution.

Examining Figure 1, one may see that in general the oral
directions test gives a type of distribution which is approx-
imately similar to the Gaussian normal distribution shown
in Figure 2. Our distribution is limited slightly at the lower
end, and it is easy to imagine that the introduction of two

8 AMERICAN INTELLIGENCE

or three items easier than any in the present test would
give us a step-down at the zero end of the scale similar to
the one at the upper end.

The oral directions test really gives an excellent score
distribution. It is not a "speed" test in the popular sense,
for the time limits for each item, while short, give ample
time for following the directions. If a person understands
the directions he can execute them easily in the time al-
lowed. If the directions are not understood, an hour to
execute them is no more generous than five seconds. In
practice this test was useful in "acclimating" the recruit to
the conditions of the examination. It was probably one of
the poorest tests in alpha as a genuine test of intelligence,
but it served its purpose as a "warming up" test. It is an
adaptation of a type of test that has been used in psycho-
logical laboratories for many years with rather mediocre
results.

Alpha Test 2. ^Arithmetical Problems
Time limit : 5 minutes

Test 2 is more of a reasoning test than a measure of
proficiency in the fundamental arithmetical operations.
The first items really constitute a literacy test, for if a
person can read, he can answer the questions correctly.
The distribution of scores in this test is shown in Figure 3.
The zero scores (66 in number) are probably due to the
inclusion of illiterates in the group of 1047 cases. Disregard-
ing the zero scores, the distribution is regular. This test
illustrates admirably the principle of fixing a time limit
such that very few people can answer all the items cor-
rectly. The approximate rule adopted in fixing the time
limit in the first instance was that this limit should be such
that not more than five per cent, of an unselected group

Get the answers to these examples as quickly as you can.
Use the side of this page to figure on if you need to.

{1 How many are 5 men and 10 men? Answer ( 15
2 If you walk 4 miles an hour for 3 hours, how far
do you walk? Answer ( 12

1 How many are 60 guns and 5 guns? , Answer (

2 If you save $9 a month for 3 months, how much will you
save? Answer (

3 If 48 men are divided into squads of 8, how many squads will
there be? Answer (

4 Mike had 11 cigars. He bought 2 more and then smoked 7.
How many cigars did he have left?. Answer (

5 A company advanced 8 miles and retreated 2 miles. How far
was it then from its first position? Answer (

6 How many hours will it take a truck to go 42 miles at the rate

of 3 miles an hour? > .Answer (

7 How many pencils can you buy for 60 cents at the rate of 2

for 5 cents? Answer (

8 A regiment marched 40 miles in five days. The first day they
marclied 9 miles, the second day 6 miles, the third 10 miles, the
fourth 6 miles. How many miles did they march the last
day? , Answer (

9 If you buy 2 packages of tobacco at 8 cents each and a pipe for
65 cents, how much change should you get from a two-dollar
bill? , , . , Answer {

10 If it takes 4 men 3 days to dig a 120-foot drain, how many men

are needed to dig it in half a day? . .Answer (

11 A dealer bought some mules for $2,000. He sold them for
$2,400, making $50 on each mule. How many mules were
there? Answer (

12 A rectangular bin holds ^00 cubic feet of lime. If the bin is

10 feet long and 5 feet wide, how deep is it? Answer (

13 A recruit spent one-eighth of his spare change for post cards
and twice as mugh for a box of letter paper, and then had $1 .00
left. How much money did he have at first Answer (

14 If 3H tons of clover cost $14, what will 6J^ tons cost?. .Answer \

15 A ship has provisions to last her crew of 700 men 2 months.
How long would it last 400 men? Answer (

16 If an aeroplane goes 250 yards in 10 seconds, how many feet
does it ^o in a fifth of a second? , Answer (

17 A U-boat makes 8 miles an hour under water and 20 miles on
the surface. How long will it take to cross a 100-mile channel,

if it has to go two-fifths of the way under water?. Answer (

18 If 134 squads of men are to dig 3,618 yards of trench, how
many yards must be dug by each squad? .Answer (

19 A certain division contains 5,000 artillery, 15,000 infantry, and
1,000 cavalry. If each branch is expanded proportionately
until there are in all 23,100 men, how many will be added to the
artillery? .Answer (

20 A commission house which had already supplied 1 ,897 barrels
of apples to a cantonment delivered the remainder of its stock
to 37 mess halls. Of this remainder each mess hall received 54
barrels . What was the total number of barrels supplied? . . Answer C

Plate II. Alpha Test 2 : Arithmetical Reasoning (Form 8).

AMERICAN INTELLIGENCE

would complete all the items in a test. In our group of 1047
cases, 5 persons answered 18 problems correctly in the five
minutes allowed, but no one answered more than 18 prob-
lems correctly. Of course no one was expected to answer
them all. If a person passed Q5% of each test in alpha he
was graded "A"; perfection was not required.

TEST 2, ALPHA

100

Figure 3. Distribution of scores of the Arithmetical Reasoning
test. (From p. 624, Memoir XV.)

One often hears the statement that the army tests were
"speed" tests, and penalized the slow but accurate indi-
vidual. Experiments were made to determine how the re-
sults would change with extended time. A group of 475
men examined showed in Test 2 an improvement from an
average of 8.00 to 9.16 with double time. In five minutes
they solved on an average 8 problems correctly, in ten min-
utes 9.16. The relationship between single time and double
time scores may be measured by the statistical value known
as the coefficient of correlation. Two measures that stand
in a perfect one to one correspondence have a coefficient

AMERICAN INTELLIGENCE 11

of correlation of 1.00. Two measures that stand in a perfect
chance relationship have a correlation coefficient of 0. In
practice it is found that a correlation of 0.90 is so high that
one might substitute one series of measures for the other
without seriously changing the results. The correlation be-
tween the single time and double time scores was 0.937, a
value so high that it indicates that there were very few
changes in the relative position of the members of the
group, and that such changes as occurred were small.

The experiments that were conducted on time limits with
the various tests all pointed to the conclusion that the re-
sults would not be changed with the more extended time
limits. Of course the absolute scores would be higher with
the extended time, but the relative position of the mem-
bers of the group would be about the same. In the experi-
ment on double time referred to above, all the tests from 2
to 8 in alpha showed coefficients of correlation between
single time and double time above 0.90 except Test 3
(0.879), and the correlation of the two total scores ob-
tained under single and double time was 0.967. The army
experimenters after considering all the evidence concluded
that "doubling the time does not result in any demonstrat-
able improvement in alpha as a whole." (p. 417). It is prob-
ably true that very high scores depend on "speed," but
inasmuch as a person only needed to answer correctly Q5%
of the items to be rated "A" and 50% of the items to be
rated "B," it can not be considered that "speed" is a factor
that would affect the results seriously.

The army findings of a correlation of 0.967 between the
single time and double time trials of alpha, and the general
conclusion that the results would not have been changed
appreciably with more liberal time allowances, definitely
controvert the popular belief that anything which is per-
formed with a time limit handicaps the "slow but sure"

12 AMERICAN INTELLIGENCE

individual. Popular judgment classifies the population into
two groups, the "slow but sure" and the "quick and inac-
curate," and would have us believe that the quick type
must of necessity be inaccurate, and that the sluggish indi-
vidual is infallible. Science shows us that we really rate
individuals on two scales, a scale of speed and a scale of
accuracy, and that we find people who are both quick and
accurate as well as people who are slow and inaccurate.
Science would elaborate the popular classification by add-
ing these two types. The popular "slow but sure" charac-
terization is more apt to be an apology for dullness than
a scientific diagnosis. At least in our consideration of the
army test results we may definitely discard the opinion
that we are testing "speed" rather than intelligence. The
arithmetical reasoning test in alpha actually proved to be
one of the best tests in the series.

Alpha Test 3. Practical Judgment

Time limit : Ij^ minutes

The practical judgment test is one of the most interest-
ing tests in alpha from many standpoints. There is no other
test in alpha which contains, in all of the five forms used, so
many individual items that may be criticised by a person
who actually inquires into the logical validity of the an-
swers accepted as correct. Item 12, for instance, might
profitably be taken as the subject of an intercollegiate de-
bate, as it has been the subject of many debates in the
history of penology. The critics of the army tests are all
too apt to consider the whole scale invalid if they can dis-
cover a single incorrect item, for they fail to realize that a
person could fall down on 35% of the individual items and
still be rated "A."

Tbis is a test of common sense. Bel6w are sixteen questions. Three answers are given to each
question. You are to look at the answers carefully; then make a cross in the square before the best
answer to each question, as in the sample:

/ Why do we use stoves?

sample!

Because
thfey look well
they keep us warm
they are black

Here the second answer is the best one and is marked with a cross
on until time is called.

^ a

Begin with No. 1 and keep

1 It is wiser to put some money aside and not
spend it all, so that you may

D prepare for old age or sidcness

D collect all the different kinds of money

n gamble when you wish

2 Shoes are made of leather, because
D it is tanned

P it is tough, pb'able and warm
in it cari be blackened

3 "Why do soldiers wear wrist watches rather
than pocket watches? Because

D they keep better time
D they are harder to break
Q they are handier

4 The mam reason why stone is used for building
purposes is because

D it makes a good appearance
n it is. strong and lasting
Q it is heavy

5 Why is beef' better food than cabbage?

D it tastes betta*

n it is mo^ nourishing

it is hairdo to obtain

If some one does you a favor, what dudd you

do?

D try to forget it

n steal for him if he asks yoti to

n return the favor

If you do not get a letter from home, which you

know was written, it may be because

D it was lost in the mails

D you forgot to tell your people to write

D the postal service has been discontinued

The main thing the. farmers do i& to

Q supply luxiu-ies

n make work for the unemployed

Q feed the nation

^"00 to. No. 9 above

9 If a man who can't swim should fall into a
river, he should

O yell for help and try to scramble out
D dive to the bottom and crawl out
O lie on his back and float

10 Glass insulators are used to fasten telegraph
wires because

O the glass keeps the pole from being burned
D the glass keeps the current from escaping
n the glass is cheap and attractive

11 If your load of coal gets stuck in the mud,
what should you do?

D leave it there

n get more horses or men to pull it out

D thro^ off the load

12 Why are criminals locked up?
O to protect society

n to get even with them
D to make them work

13 Why should a married man have: his life in
snud? Because

n death may come at any time

n insurance companies are usually honest

D his family will notihen suffer iJE he dies

14 In Leap Year February has 29 days because
D February is a short month

D some people are bom on February 29th
n otherwise the calendar would not come out
right

15 If you are held up and robbed in a strange city,
you should

D apply to the police for help

P ask the first man you meet for money ta

get home
D borrow some money at a bank

16 Why should we have Congressmen? Because
n the people must be ruled

D it insures truly representative government
D the people are too many to meet and make
their laws

Plate III. Alpha Test 3: Practical Judgment (Form 8).

14 AMERICAN INTELLIGENCE

The distribution of the scores made in Test 3 is shown
in Figure 4.

TEST 3, ALPHA

150-

100-

Figure 4. Distribution of scores of the Practical Judgment test.
(From p. 624, Memoir XV.)

Disregarding the large number of zero scores (163), which
are probably the result of illiteracy plus failure to under-
stand instructions, and also recognizing the fact that a few
low positive scores may be due to chance, we may look at
the distribution as entirely satisfactory.

Many persons object to the short time limit {V/^ min-
utes), but the test was undoubtedly more effective with this
short limit than it would have been with the longer time.

AMERICAN INTELLIGENCE 15

The average score improves from 6.32 to 9.85 with double
time, and the correlation between Ij^ and 3 minutes work
on the test is 0.879. There are decided indications that
double time would not be useful in improving the record
of those whose score was high in the first 1)^ minutes.

A very excellent criterion of the efficiency of a test is
its value in differentiating between officers and men. In
general, a sample of officers would contain a larger per-
centage of men of high intelligence than a sample of en-
listed men. The amount that a test differentiates the groups
would indicate the value of the test. This test of practical
judgment was the worst test in the whole series in differ-
entiating officers from men. If we used this criterion alone
there would be no possible excuse for retaining the test in
the series. In differentiating officers from men, it was about
twice as bad as the next to the poorest test (oral direc-
tions).

On the other hand, we need tests in alpha which are
effective at the lower end of the scale, and we can set up
as our criterion here the value of the test in differentiating
between feeble-minded individuals and enlisted men. The
alpha tests were given to the high grade feeble-minded
population of two custodial institutions, and the results
compared with a group of 300 English speaking enlisted
men. Test 3 proved to be very much superior to any other
test in the series in differentiating between feeble-minded
individuals and enlisted men. This fact more than justifies
the inclusion of Test 3 in the scale.

All of these facts are difficult to interpret. My own inter-
pretation is that the sixteen items did not measure or grade
"practical judgment" in any sense, but that the inclusion
of at least one very obviously false and really quite silly
alternative in each item acted as an effective pitfall for the

16 AMERICAN INTELLIGENCE

feeble-minded. At least we are sure that the actual experi-
mental results are conclusive enough to dispose of any and
all arm-chair criticisms.

Alpha Test 4. Synonym-Antonym

Time limit : Ij^ minutes

If one will review the experimental literature of intelli-
gence testing, he will find the synonym-antonym or "op-
posites" test, used sometimes as a test of controlled asso-
ciation, sometimes as a test of vocabulary, sometimes as a
test of intelligence, but uniformly with excellent results.
Given a group with a knowledge of English and sufficient
intelligence to understand the nature of the problem,
the synonym-antonym test will give as good a differentia-
tion between the bright and dull members of the group,
rated by an outside criterion, as any other standard test
now available.

The distribution of the scores in Test 4 is shown in
Figure 5.

The most striking feature of the distribution is the large
number of scores that were either zero or one (393). This
large number of zero scores is due to three causes. First,
the illiterate group could not attempt it. Second, the stupid
and literate could not understand the instructions and
could not make the kind of judgment demanded. Third, in
the long run chance or random responses would give scores
around zero, for in scoring all tests that were a 50-50
guess, the total score was the number of right responses
minus the number of wrong responses. If a person under-
lined ^'same^' for every item, his score would be 20 right
minus 20 wrong, or zero. If he merely guessed, he would,

400-

350-

300-

250

200-

150-

100-

50-

TEST 4, ALPHA

Figure 5. Distribution of scores of the Synonym-Antonym Test.
(From p. 625, Memoir XV.)

If the two words of a pair mean the same or nearly the same, draw a
line under same. If they mean the opposite, or nearly the opposite, draw a
line under opposite. If you cannot be sure,. guess. The two samples are
already marked as they should be.

SAMPLES i ^^ ^^^ same opposite

V little small same opposite

1 no ^yes same opposite 1

2 day night same opposite 2

3 go leave same opposite 3

4 begin commence same opposite 4

6 bitter sweet same opposite 5

6 assume suppose same opposite 6

7 command obey same opposite 7

8 tease plague same opposite 8

9 diligent industrious same opposite 9

10 corrupt honest same opposite 10

11 toward from- same opposite 11

12 mascuUne feminine same opposite 12

13 complex simple same opposite 13

14 sacred hallowed same opposite 14

15 often seldom same opposite 15

16 ancient modem same opjwsite 16

17 enormous gigantic same opposite 17

18 confer grant same opposite 18

19 acquire lose same opposite 19

20 compute calculate .same opposite 20

21 defile purify . . same opposite 21

22 apprehensive ^fearful same opposite 22

23 sterile fertile same opposite 23

24 chasm abyss same opposite 24

25 somber ^gloomy same opposite 25

26 vestige trace.. same opposite 26

27 vilify praise same opposite 27

28 finite limited same opposite 28

29 contradict corroborate same opposite 29

30 immune susceptible same opposite 30

31 credit debit same opposite 31

32 assiduous diligent same opposite 32

33 transient ^permanent same opposite 33

34 paUiate mitigate same opposite 34

35 execrate revile ..same opposite 35

36 extinct extant same oppositfe 36

37 pertinent relevant same opposite 37

38 synchronous simultaneous.. . .same opposite

39 supercilious disdainful same opposite

40 abstruse recondite same opposite 40

Plate IV. Alpha Test 4: Synonym-Antonym (Form 8).

AMERICAN INTELLIGENCE 19

in the long run, guess half of the responses right and half
wrong. Chance scores would then be zero (which includes
all minus scores) and 1 or 2 on the positive side.

In general we may interpret Test 4 as a "high grade"
test. It is too difficult to give any differentiation between
low grade individuals, but it effectively grades the higher
orders of intelligence. The time limit is not too short, for
doubling the time only raises the average score from 10.50
to 12.60, and the correlation between regular and extended
time is 0.940. It is one of the most effective tests in the scale
for differentiating officers from enlisted men, and for dif-
ferentiating feeble-minded from enlisted men. The only
criticism is that it was too hard for a large number of
people examined. Figure 5 really gives only about half of
the normal distribution. If the test were so easy that the
lower end of the scale could be extended to about 20, the
distribution would become normal.

Alpha Test 5. ^Disarranged Sentences
Time limit : 2 minutes

This test is an adaptation of a type of test which gives
excellent results in the Binet-Simon scale. As it stands in
alpha it is not a particularly good test. The distribution of
scores shown in Figure 6 indicates a pile-up of zero scores
due probably to the same three causes described as op-
erating in Test 4. The test is fairly good in differentiating
between officers and enlisted men, but for some reason or
other it is the very worst test in the whole series in dif-
ferentiating between feeble-minded and enlisted men. On
the whole it is one of the poorest tests in our measuring
scale.

250 n

200

150-

100-

50-

TEST 5, ALPHA

Figure 6. Distribution of scores of the Disarranged Sentence Test.
(From p, Q2Q, Memoir XV.)

The words A EATS COW GRASS m that order are mLxed up and
don't make a sentence; but they would make a sentence if put m the-
right order: A COW EATS GRASS, and this statement is true.

Again, the words HORSES FEATHERS BAYE. ALL would make
a sentence if put in the order ALL HORSES HA\'E FEATHERS,
but this statement is false.

Below are twenty-four mixed-up sentences. Some of them are true
and some are false. When I say "go," take these sentences one at a
tkne. Think what each would say if the words were straightened out,
but don't write them yourself. Then, if what it would say is true, draw
a line under the word "true"; if what it would say is false, draw a line.
vmder the word "false." If you can not be sure, guess. The two
samples are already marked as they should be. Begin with No. 1
and work right down the page until time is called.

( a eats cow grass true, .false

SAMPLES^ ^

V n

feathers have all true'. : false

1 oranges yellow are true . .false 1

2 hear are with to ears. true, .false 2

3 noise cannon never make a true, .false 3

4 trees in nests build birds tpie. .false 4

5 oil water not and will mix , true, .fake 5

6 bad are shots soldiers all true.. false 6

7 fuel wood are coal and for used true, .false 7

8 ; moon earth the only from feet twenty the is true . . false 8

9 to life w:ater is necessary true, .false 9

10 are clothes all made cotton of true .. .false 10

11 horses automobile an arfe than slower true, .false 11 /

12 tropics is in the produced rubber true . . false 12

13 leaves the trees in lose their fall true . . false 13

14 place pole is north comfortable a the true . . false 14

15 sand of made bread powder and is true . .false 15

16 sails is steamboat usually by propelled a true . . false 16

17 is the salty in water all lakes true, .false 17

18 usually judge can we actions man his by a true . . false 18

19 men misfortune have good never true . .false 19 .

20 tools valuable is for sharp making steel true, .false 20

21 due sometimes calamities, are accident to true . . false 21

22 forget trifling friends grievances never true, .false 22

23 feeling is of painful esaltation the true . . false 23

24 begin a and apple acorn ant words with the true . . falsa 24

Plate V. Alpha Test 5: Disarranged Sentences (Form 8),

2 AMERICAN INTELLIGENCE

Alpha Test 6. ^Number Series Completion

Time limit : 3 minutes

This test is the only one in alpha demanding a high order
of intelligence almost entirely independently of the use of
language. The greatest difficulty was experienced with the
instructions for this test, when the first trial was made at
the army camps. The preliminary forms contained only two
rows of samples, and the instructions included the rather
involved statement: "In the lines below, each number is
gotten in a certain way from the numbers coming before it.
Study out what this way is in each line and then write in
the space left for it the number that should come next.
The first two lines are already filled in as they should be."
In the final alpha revision, four samples were included, and
the instructions were simplified verbally and read very
slowly. The instructions were given as follows: "Look at
the first sample row of figures at the top of the page : 2, 4,
6, 8, 10, 12; the two numbers that should come next are,
of course, 14, 16," etc., for each sample. Long pauses fol-
lowed the reading of each sample.

The distribution of responses given in Figure 7 shows
that the simplified instructions gave very good results, for
although there were many zero scores in our experimental
group of 1047 cases (244), there were probably no more
zero scores than might have been expected when we con-
sider that the mere understanding of what was wanted re-
quired considerable intelligence. On the whole the number
series completion test proved to be entirely satisfactory.

TEST 6, ALPHA

150-

100-

50-

Figure 7. Distribution of scores of the Number Series Completion
Test. (From p. 624, Memoir XV.)

f 4 6 8 10 12

9 8 7 6 5 4

SAMPLES ^2233 4 4

17 2 7 3 7,

Look at each row of numbers below, and on the two dotted lines
write the two numbers that should come next.

3 4 5 6 7 8

8 7 6 5 4 3

10 15 20 25 30 35

9 9 7 7 5 5

3 6 9 12 15 18

8 16 14 1

5 9 13 17 21 25

8 9 12 13 16 17

27 27 23 23 19 19

1 2 4 8 16 32

19 16 14 11 9 6

11 13 12 14 13 15

2 3 5 8 12 17

18 14 17 13 16 12

29 28 26 23 19 14

20 17 15 14 11 9

81 27 9 3 1 Yz

1 4 9 16 25 36

16 17 15 18 14 19

3 6 8 16 18 36

Flate VI. Alpha Test 6: Number Series Completion (Form 8).

AMERICAN INTELLIGENCE 25

Alpha Test 7. ^Analogies

Time limit : 3 minutes

The analogies test gave results on a par with the syn-
onym-antonym test. The distribution of scores given in
Figure 8 shows a large number of zero scores (284), but this
number is not larger than might have been expected, con-
sidering the amount of intelligence necessary even to
understand the nature of the task to be performed. This
test again shows only a partial distribution. If it had been
easier, it would probably have shown a symmetrical dis-
tribution.

"TEST 7. ALPHA

250 -

200-

150-

100-

50-
0--

1 1 1 1 [ 1 1 1 1 1 1 1 1 1 1

Figure 8. Distribution of scores of the Analogies Test. (From p.
625, Memoir XV.)

( sky blue : : grass table green warm Ijig

SAMPLES < fish swims:; man paper time walks girl

( day night : : white red black clear pure

In each of the liiies below* the first two words are related to each other in some way. What
you are to do in each line is to see what the relation is between the first two Words, and under-
Kne the word in heavy type that is related in the same way to the third word. Begin with
No. 1 and mark as many sets as you can before time is called.

1 shoe foot :: hat kitten head knife penny 1

2 pup dog :: lamb red door sheep book... ,..,. 2

3 spring summer :: autumn winter warm harvest rise 3

4 devil angel : : bad mean disobedient defamed good. ,. , 4

5 finger hand : : toe body foot skin nail 5

6 legs frog : : wings eat swim bird nest 6

7 chew teeth : : smell^ sweet stink odor nose 7

8 lion roar:: dog drive pony bark harness 8

9 cat tiger:: dog wolf bark bite snap 9

10 good^ad : : long tall big snake short... 10

11 giant large : : dwarf jungle small beard ugly. 11

12 winter season : : January February day month Christmas . . 12

13 skating winter :: swimming diving floating hole summer 13

14 blonde light :: brunette dark hair brilliant blonde 14

15 love friend : : hate malice saint enemy dislike 15

16 egg bird : : seed grow plant crack germinate 16

17 dig trench :: build run house spade bullet 17

18 agree quarrel i : friend comrade need mother enemy 18

19 palace king : : hut peasant cottage farm city 19

20 cloud-burst shower ; : cyclone bath breeze destroy West. . . 20

21 Washington ^Adams:: first president second last Bryan .. 21

22 parents command : : children men shall women obey 22

23 diamond rare:: iron common silver ore steel l . 23

24 yes affirmative :: no think knowledge yes negative 24

25 hour day:: day night week hour noon 25

26 eye head : : window key floor room door 26

27 clothes man : : hair horse comb beard hat 27

28 draw picture :.: make destroy table break hard. 28

29 automobile wagon : : motorcycle ride speed bicycle car 29

30 granary wheat :: library- read books paper chairs 30

31 Caucasian English : : Mongolian Chinese Indian negro yellow. . 31

32 Indiana United States : : part hair China Ohio whole 32

33 esteem despise :: friends Quakers enemies lovers men 33

34 abide stay : : depart come hence leave late 34

35 abundant scarce :: cheap buy costly bargain nasty 35

36 whale large : : thundei loud rain lightnmg kill 36

37 reward- hero : : punish God everlasting pain traitor .... 37

38 music soothing :: noise hear distrdcting sound report. ...... 38

39 book writer :: statue sculptor liberty picture state 39

40 wound pain : : health sickness disease exhilaration doctor. . 40

Plate VII. Alpha Test 7: Analogies (Form 8).
96

AMERICAN INTELLIGENCE 27

In the construction of this test it was dehberately plan-
ned to make many items very difficult by introducing as a
wrong alternative a word that was frequently associated
with the key word. For instance, we find in the last 20 items
the following easily associated pairs which would be wrong :
(21) first-last, (24) no-yes, (25) day-night, (26) window-
door, (27) hair-comb, (28) make-break, (29) motor-cycle-
ride, (30) library-read, (32) part-hair, (35) cheap-buy, (36)
thunder-Kghtning, (38) noise-hear, (39) statue-Kberty, (40)
health-sickness. The test therefore involves not only the
selection of the right word, but the refusal to accept as the
solution a word that is exceedingly attractive owing to fre-
quent associations.

The analogies test is the most effective test in the entire
series in differentiating officers from men. For some reason,
not understood, it does not rank high in differentiating
feeble-minded from enlisted men. The scores show a con-
siderable average improvement with extension of time limit
(8.60 to 12.46) but a correlation of 0.920 between three and
six minutes work. On the whole it is an excellent test.

28 AMERICAN INTELLIGENCE

Alpha Test 8. Information

Time limit : 4 minutes

The army information test has been criticised more than
any other test, and it has undoubtedly received much more
abuse than it deserves. From the standpoint of test con-
struction it is satisfactory, for, aside from the zero scores
probably due to illiteracy, the distribution as shown in
Figure 9 is rather good.

150-1

100-

TEiST 8, ALPHA

{ 1 I 1 I M I I i I- I 1 I i I i M I r
10 20 30 40

Figure 9. Distribution of scores of the Information Test. (From
p. 626, Memoir XV.)

The most frequent charge made against the test is that
a person could fail in certain items and still be intelligent.
This is certainly true, and the criticism would be valid if
anyone were expected to answer all the items, or if he were
considered unintelligent if he failed. The average person
answered less than 15 items correctly.

Notice the sample sentence:

People hear with the eyes ears nose mouth

The correct word Is ears, because it makes the truest sentence.

In each of the sentences belo^ you have four choices for the last word. Only one of thenx is cor-
rect. In each sentence draw a line imder the one of these four words which makes the truest sentence.
If you can not be sure, guess. The two samples are already marked as they should be.

People hear with the eyes ears nose mouth
SMIPLES^ "^

France is b Europe Asia Africa Australia

1 The apple grows on a shrub vine bush tree 1

2 Five hundred is played with rackets pins cards dice ,.... 2

3 The Percheron is a kind of goat horse cow sheep , 3

4 The most prominent industry of Gloucester is fishing packing brewing automobiles.. 4

5 Sapphires are usually blue red green yellow ..., 5

6 The Rhode Island Red is a kind of horse granite cattle, .fowl 6

7 Christie Mathewson is famous as a writer artist baseball player comedian 7

8 Revolvers are made by Swift Sc Co. Smith & Wesson W. L. Douglas B. T. Babbitt. 8

9 Carrie Nation is known as a singer temperance agitator sufffftgist nurse 9

10 "There's a reason" is an "ad" for a drinJt revolver flour cleanser 10

^11 Artichoke is a kind of hay corn vegetable fodder H

12 Chard is a fish lizard vegetable snake -,..... li

13 Cornell Uriversity is at Ithaca Cambridge AnnapoUs New Haven. .., 13

14 Buenos Aires is a city of Spam Brazil Portugal Argentina 14

15 Ivory is obtained from elephants mines oysters reefs ^^ 15

16 Alfred Noyes is f anlous as a painter poet musician sculptor 16

17 The armadilid is a kind of ornamental shrub animal musical instrument dagger 17

18 The tendon of Achilles is in the heel head shoulder abdomen 18

19 Crisco is a patent medicine disinfectant tooth-paste food product 19

20 An aspen is a machme fabric tree drink 20

21 The sabre is a kind of musket sword cannon pistol ;.. , 21

22 The mimeograph is a kind of typewriter copying machine phonograph pencil 22

23 Maroon is a food fabric drink color 23

24 The clarionet is used in music stenography ' book-bmdmg lithography , 24

25 Denim is a dance food fabric drink , 25

26 The author of "Huckleberry Fmn" is Poe Mark Twain Stevenson Hawthorne 26

27 Faraday was most famous m Uterature war religion science 27

28 Air and gasolene are mixed in the accelerator carburetor gear case differential ' 28

29 The Brooklyn Nationals are called the Giants Orioles Superbas Indians .' 29

30 . Pasteur is most famous in poUtics literature war science ... . 30

31 Becky Sharp appears in Vanity Fair Romola The Christmas Carol Henry IV 31

32 The number of a Kaffir's legs is- two four six- eight , 32

33 Habeas corpus is a term used in medicine law theology pedagogy 33

34 Ensilage is a term used in fishing athletics farming hunting , 34

35 The forward pass is used in tennis hockey football golf 35

36 General Lee surrendered at Appomattox in 1812 1855 1886 1832 36

~^ 37 The watt is used in measuring wind power rainfall water power electricity 37

38 The Pierce Arrow car is made in Buffalo Detroit Toledo Flint ,. 38

'39 Napoleon defeated the Austrians at Friedland Wagram Waterloo Leipzig 39

40 An irregular four-sided figure is called a ,5cholium triangle trapezium peirtagon 40

Plate VIII. Alpha Test 8: Information (Form 8),

1 ^

30 AMERICAN INTELLIGENCE

The test was devised to sample as many fields of infor-
mation as it was possible to sample with 40 items. In gen-
eral the five information tests in the five forms of alpha
sampled similar fields. For instance, the advertising slogans
which appear in the five forms are "Hasn't scratched yet"
(Form 5), "The makings of a nation" (Form 6), "Even-
tually, why not now?" (Form 7), "There's a reason" (Form
8), and "The flavor lasts" (Form 9), while the Overland,
Buick, Rolls-Royce, Pierce Arrow and Packard appear in
each of the five forms.

Information tests vary considerably in construction.
There is a great difference between asking for the date of
Lee's surrender and asking a person to choose between the
four dates, 1812, 1865, 1886 and 1832. And again, a person
is merely asked to elect whether the bassoon, xylophone,
cymbal, clarionet and piccolo, appearing in each of the ^ve
forms of alpha, should be used in music, stenography, book
binding, or lithography. Approximately a third of the times
test for vocabulary rather than information in the literal
sense. If a person, for instance, knows what a Zulu, or a
Korean, or a Hottentot, or a Kaffir or a Papuan is, he very
obviously knows the number of his legs.

As a rule women object to the information test more than
men because the test samples rather heavily the fields of
sport, mechanical interests, etc. The chances are that this
test would penalize women rather heavily, but as a general
rule the results of comparing the two sexes on alpha as a
whole at various colleges show very slight differences in
favor of the men. The sex differences found are not large
enough to be significant.

At Camp Lee a group of 164 captains and 200 enlisted
men of the same general intelligence level (i. e. A and B)
were examined. The test showing the greatest differentia-
tion between the two groups was the information test. The

AMERICAN INTELLIGENCE 31

only other test showing a difference in favor of the captains
was Test 4 (synonym-antonym test), while the test for arith-
metical reasoning (Test 2) and practical judgment (Test 3)
showed differences in favor of the A and B enlisted men.
The differences were, of course, small, but the greatest dif-
ference was shown by the information test. In differentiat-
ing between officers and the general run of enlisted men,
the information test was fairly effective, and it was very
nearly as good as the arithmetical reasoning and synonym-
antonym test in differentiating between feeble-minded in-
dividuals and enlisted men, in spite of the fact that the
feeble-minded obtained a somewhat higher percentage of
their total score from this test than did the enlisted men.
After weighing all the evidence, it would seem that we
are justified in ignoring most of the arm-chair criticisms of
this test and in accepting the experimental evidence tend-
ing to show that the test was a fairly good one. The assump-
tion underlying the use of a test of this type is that the
more intelligent person has a broader range of general in-
formation than an unintelligent person. Our evidence shows
that this assumption is, in the main, correct.

SECTION II
EXAMINATION BETA

When we turn to examination beta, we meet an en-
tirely different problem, that of testing the inteUigence of
wholly or partially illiterate persons who could not take
alpha on account of their language handicap, of testing non-
English speaking persons some of whom knew only the
simplest commands in English, and low grade individuals
who did not have sufficient intelligence to make a substan-
tial score on alpha. At the time of the first try-out of the
army tests in the fall of 1917 at four cantonments, Devens,
Dix, Lee and Taylor, examination a, the fore-runner of
alpha, was in use, and various types of individual examina-
tions were being tried out, but there was no non-verbal
group test. To meet this need, a preliminary try-out of fif-
teen tests was made early in 1918, and a final examination
composed of seven tests was subsequently published and
widely used throughout the country. In the following pages
the seven beta tests are reproduced in Plates 9 to 15, the
method of administering them is described, and the results
from the experimental group of 1047 cases are presented.
Examination beta was given under the most rigid experi-
mental conditions. The experimenter stood on a platform
back of which was a large black-board on which small dup-
licates of the seven tests could be shown one at a time. A
demonstrator, whose duties were to act out the test prob-
lems on the black-board, was an essential part of the ex-
periment. The experimenter in pantomime showed the de-
monstrator what to do on the black-board, then, after his

AMERICAN INTELLIGENCE

performance was completed, orderlies throughout the room
explained to the men that they were to go ahead with the
test and do what the demonstrator had done. The orderlies'
vocabulary was limited to "Yes," "No," "Sure," "Good,"
"Quick," "Hurry up," "How many?" "Same," "Do it,"
"Fix it." The experimenter used just as few words as pos-
sible, and acted out every spoken sentence by pointing,
motioning, etc. The demonstrator never spoke. His duties
consisted of doing before the group just what the^group
was expected to do with the examination blanks.

Beta Test 1. Maze
Time limit : 2 minutes

The black-board was turned so that two sample mazes
as shown in Figure 10 appeared. The experimenter traced

TEST 1

i-j^j=4rj^-

1 ^

Figure 10. Black-board chart for demonstrating the Maze Test.

AMERICAN INTELLIGENCE

through the first maze on the black-board, and then mo-
tioned the demonstrator to go ahead. The demonstrator
traced through the maze with crayon very slowly. The ex-
perimenter then traced through the second maze and mo-
tioned the demonstrator to go ahead. The demonstrator
in tracing this maze made a mistake by crossing the line
at the end of a blind alley, was corrected by the experi-
menter with vigorous shakes of the head and "no-no," and
made to re-trace his path back to where he could start right
again. The demonstrator then traced through the rest of

TEST

1, BETA

200-

150-

100-

J"^

50-

U ""

(

,' ' '

1 1 i

-T

Figure 11. Distribution of scores of the Maze Test. (From p. 627,
Memoir XV.)

T r

u >

lAl

4 -^ -.

PZa^e IX. Beta Test 1 : Maze.
35

AMERICAN INTELLIGENCE

the maze with great semblance of haste, stopping momen-
tarily at each ambiguous point only. The experimenter then
motioned to the group to do the same thing on their exam-
ination blanks. The experimenter and the orderlies walked
about the room, motioning to the men who were not work-
ing, and saying, "Do it, do it, hurry up, quick."

The results of the maze test are shown in Figure 11. The
difference' between the distribution of scores of this test
and the alpha tests is remarkable. In the first place, our
large number of zero scores has disappeared only 19 in
our group of 1047 failed to make any score. In the second
place, the test is entirely too easy, for it is apparent that
the men in the upper end of the scale could have done more
in the time allowed. The maze test was intentionally made
easy in order to get everybody started. We have at last
found a test in which practically everybody can do some-
thing. Aside from the language involved, every test in alpha
is harder than this beta maze test, for no alpha test has less
than 7% zero scores.

Beta Test 2. Cube Analysis
Time limit : %}/2 minutes

The black-board was turned to show a series of cubes
like that in Figure 12. On a shelf was a real three cube

TEST 2

Figure 12. Blackboard chart for demonstrating the Cube Analy-
sis Test.

AMERICAN INTELLIGENCE

model similar to the first one on the black-board. The ex-
perimenter pointed to the three cube picture on the black-
board, then to the model on the shelf, then to the picture
on the black-board, and asked, "How much?" The experi-
menter then counted aloud, putting up his fingers while
counting, and encouraged the men to count with him. The
experimenter then tapped each cube on the black-board
and asked the demonstrator, "How much.f^" The demon-
strator then went to the black-board, counted the cubes by
pointing, and wrote the number 3 in the space below the
illustration. A similar performance was enacted for the
other three problems on the ^black-board, the models being
shown and elaborately counted.

The distribution of scores of the cube analysis test is
shown in Figure 13. Here we find a somewhat larger num-

100-

50-

TEST 2, BETA

Figure 13. Distribution of scores of the Cube Analysis Test.
(From p. 627, Memoir XV.)

ber of zero scores (54) than in the maze test, but a fairly
good distribution in general. On the whole the test is easy.

y ^ y A

I ffl

y y ./

^5f1

Plate X. Beta Test 2 : Cube Analysis.

AMERICAN INTELLIGENCE 39

Beta Test 3. X-0 Series

Time limit : 1% minutes

The black-board was turned to show the chart repro-
duced in Figure 14. The experimenter traced with a pointer

TES T 3

goioioioi 1 1 1 1

ixi ixi ixi ixi 1 1 n

loioixioiohdoioixi 1 1 1 1

ixixixioixiqxixixlolxl

Figure 14. Black-board chart for demonstrating the X-0 Series
Test.

each "O" in the top chart, and then wrote (with his pointer)
an imaginary "O" in the four remaining spaces. The dem-
onstrator then filled in the four "O's" with crayon. The
experimenter then traced the first "X" by tracing a semi-
circle above the chart and so on. The demonstrator then
filled in every other space with an "X" following the ex-
perimenter's elaborate exercise. The demonstrator then
worked out the remaining problems with the same ritual,
following which the men were instructed to go ahead.

The distribution of scores of the X-0 series test is shown
in Figure 15.

The army writers state that: "Beta 3 defies interpreta-
tion." (p. 638.) We know that the test was devised to dup-
licate in pictorial form the number series completion test

AMERICAN INTELLIGENCE

150-

100-

TEST 8, BETA

I I I I i I I I ,1 I I

5 10

Figure 15. Distribution of scores of the X-0 Series Test. (From
p. 627, Memoir XV.)

in alpha. Aside from knowing the purpose of the test, we
know very little about it. My own guess would be that the
first five or six items were entirely too easy, and that if
they had been disregarded in the scoring, as practice items,
and six more items added, comparable in difficulty to the
last six items, the distribution would have been satisfactory.

Ixlxlxl

X X

OXX

x]x

ox xox x.x No

Ai 11

xHo|xl<,|x|xloU|

U |X|0|X|X|0|X|X|X|<>|X|0|X|X|0|X|^|X|C| I I I I I I I I I

lg |X|X|X|X|o|o|o|x|X|.|x|X|x|X|.|o|o|x|x|o| I I I I I M I I I
Plate XI. Beta Test 3 : X-O Series.

42 AMERICAN INTELLIGENCE

Beta Test 4. Digit-Symbol
Time limit : 2 minutes

The black-board was turned to the chart shown in Fig-
ure 16. The experimenter pointed to each number and then

TEST 4

3ll2|3|2l|2|l|3|4|7 5l4lte^

III IN

Figure 16. Black-board chart for demonstrating the Digit-Symbol
Test.

to the symbol under it. The experimenter then pointed to
the number 3 in the sample, then to the space below it, then
to the number 3 in the key above, then to the symbol for
3, and finally traced the outline of the symbol for 3 in the
proper space in the sample. This procedure was then re-
peated for the first ^yq samples. The demonstrator then
went to the black-board, and worked through the process
of filling in the symbols under the figures, touching each
figure and symbol in the key while he drew the proper sym-
bol in the sample. The group then proceeded to fill in the
symbols on the test blank.

The distribution of scores of the digit-symbol test is
shown in Figure 17.

AMERICAN INTELLIGENCE

100-

50-

TEST

4. ri BETA

^ M i M

TT-r

Figure 17. Distribution of scores of the Digit-Symbol Test. (From
p. Q'^S, Memoir XV.)

This type of distribution is the same as that given by the
eight alpha tests, the zero scores representing failure to
understand instructions, and the distribution being fairly
regular. The digit-symbol test is a standard test, and the
results in beta are entirely satisfactory.

j8^

J9_

3 2

1 4

A S

\ 7

5 4

8 1

L 9

9 5

3 7

' 4 6

7 e

"9^

4 :

I 5

7 (

> 2

1 I

^ .

Plate XII. Beta Test 4: Digit-Symbol.

AMERICAN INTELLIGENCE 45

Beta Test 5. Number Checking

Time limit : 3 minutes

The black-board was turned to the chart shown in Fig-
ure 18. The experimenter pointed first to the 6 in the left

TEST 5

62 62

59 56

327 327

249 249

1536 1536

3745 3745

45010 45001

62019 62019

Figure 18. Black-board chart for demonstrating the Number
Checking Test.

hand column, then to the 6 in the right hand column, then
to the 2 in the left hand column and to the 2 in the right
hand column, nodded his head, said, "Yes," and made an
imaginary cross on the dotted line. The demonstrator then
made an "X" on the line. The experimenter repeated the
procedure for the second pair, but indicated clearly, by
shaking his head and saying, "No," that the 9 and the 6
were not alike. The experimenter then repeated the pro-
cedure for three more sets, getting the men in the room to

AMERICAN INTELLIGENCE

say, "Yes," if the pairs were identical. The demonstrator
then worked out the remaining items.

The results of the number checking test shown in Fig-
ure 19, give the same distribution characteristic of the

TEST 5, BETA

Figure 19. Distribution of scores of the Number Checking Test.
(From p. 628, Memoir XV.)

alpha tests. The instructions were clear and the test was en-
tirely satisfactory. This test is an adaptation of a standard
test in use for many years.

650 650

041 044

2579 2579

3281 3281

55190 55102

39190 39190

658049 : . . . 650849

3295017 3290517

63015991 63019991

39007106 , 39007106

69931087 69931087

251004818 251004418

299056013 299056013

36015992 360155992

3910066482 391006482

8510273301 8510273301

263136990 263136996

451152903 451152903

3259016275 3295016725

582039144 582039144

61558529 61588529

211915883 219915883

670413822 670143822

17198591 , . . . 17198591

36482991 36482991

10243586 10243586

659012534 6590211354

388172902 381872902

631027594 631027594

2499901354 2499901534

2261059310 2261659310

2911038227 2911038227

313377752 313377752

1012938567 1012938567

7166220988 7162220988

3177628449 3177682449

468672663 468672663

9104529003 9194529003

3484657120 3484657210

8588172556 8581722556

3120166671 3120166671

7611348879 76111345879

26557239164 26557239164

8819002341 8819002341

6571018034 6571018034

38779762514 38779765214

39008126557 39008126657

75658100398 . . 75658100398

41181900726 41181900726

6543920817 6543920871

Plate XIII. Beta Test 5 : Number Checking.

48 AMERICAN INTELLIGENCE

Beta Test 6. ^Picture Completion

Time limit : 3 minutes

The black-board was turned to the chart shown in Fig-
ure 20. The experimenter pointed to the hand and said,

TEST 6

Figure 20. Black-board chart for demonstrating the Picture Com-
pletion Test.

"Fix it." The demonstrator looked puzzled. The experi-
menter pointed to the place where the finger was missing,
and said, "Fix it; Rx it." The demonstrator then drew the
finger. The experimenter then pointed to the fish, and the
place for the eye, and said,"Fix it." After the demonstrator
had drawn in the eye, the experimenter pointed to each of
the drawings and said, "Fix them all." After the demon-
strator had worked out all the remaining drawings the
group proceeded to complete the drawings in the beta blank.
The results of the pictorial completion test as given in
Figure 21 show an excellent distribution of the same general
type as the distribution of the eight alpha tests. The num-
ber of zero scores (12) on this test is smaller than that of
any other test in the entire alpha-beta series. This is an
excellent test.

100-

TEST 6, BETA

Figure 21. Distribution of scores of the Picture Completion Test.
(From p. 628, Memoir XV.)

V.y.

Plate XIV. Beta Test 6: Picture Completion.
SO

AMERICAN INTELLIGENCE 51

Beta Test 7. Geometrical Construction

Time limit : %}/2 minutes

The black-board was turned to the chart shown in Fig-
ure 22. The experimenter pointed to the square on the

TEST 7

cQn

Figure 22. Black-board chart for demonstrating the Geometrical
Construction Test.

black-board, and taking two pieces of cardboard the same
size as the drawings at the left of the square, fitted them
on the two drawings. He then fitted the two pieces of card-
board together on the square to show that they would fill
it, and motioned to the demonstrator who drew, in the
square, the lines indicating the manner in which the two
pieces would fit. This procedure was repeated for the next
two samples. The demonstrator then worked out the last
sample.

The responses given in Figure 23 show, aside from the zero
scores, a peculiar distribution, the form of which may be
interpreted by assuming that the test was too hard in its

AMERICAN INTELLIGENCE

BETA

100-

50-

Figure 23. Distribution of scores of the Geometrical Construction
Test. (From p. 628, Memoir XV.)

beginning and too easy at the end. We can picture the dis-
tribution of Test 7 as shown in Figure 23 as having come
from the middle range of a more complete test such as
Test 5, Figure 19. If we cut the distribution of Test 5 from
10 to 19, we can picture the range in which Test 7 was
working. We may assume, then, that the inclusion of a few
easier items and five or ten harder ones might have given
a distribution similar to that of the alpha tests. The test is
faulty because of the limitation of range at both ends.

^17

D
D

d
^

Plate XV. Beta Test 7 : Geometrical Construction.
53

SECTION III

THE INDIVIDUAL EXAMINATIONS

The greatest contribution of the army psychologists to
the development of mental tests was that of creating the
two group tests, alpha and beta, that have been discussed.
Methods of individual examining had been in existence for
several years. The basic series of tests of the individual ex-
amination was the Stanford Revision of the Binet-Simon
scale, which had become a standard measurement and
needs no description here. Persons interested in this method
should read Terman'si book on the Stanford scale. A
method of abbreviating the Stanford-Binet scale was
worked out in the army, and proved satisfactory.

The distribution of the scores in terms of the "mental
ages" of the 653 men in the special experimental group of
1047 cases who took the Stanford-Binet examination was
as follows:

"Mental Ages" 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

No of cases 1 2 122 62 66 69 8169 77 63 54 47 34 5

A rough inspection of these figures shows that they give
us the Gaussian normal distribution. The results obtained
from the Stanford-Binet examination may be taken as en-
tirely reliable without question.

One difficulty in the popular interpretation of the results
on the Stanford-Binet scale, and other scales constructed
on the same principle, is the unfortunate use of the term
"mental age," a term first used by Binet and subsequently

^L. M. Terman. The Measurement of Intelligence. Boston, 1916, Pp. 362.

AMERICAN INTELLIGENCE 55

used in this country. The term "mental age" has no signi-
ficance whatsoever aside from the particular scale from
which it was derived. A person might have a "mental age"
of 13 on the Stanford-Binet scale, of 11 on Goddard's
translation of Binet's 1908 scale, of 12 on Goddard's 1911
scale, and so forth for every scale in use. The term "mental
age" really means a score on a particular series of tests.
Through rather general usage, the Stanford-Binet scale is
being adopted in this country as a standard.

The Stanford-Binet scale was constructed out of some
90 different tests arranged for different age levels, six for
each age level from 3 to 10, eight for 12, six for 14, six for
16 or "average adult," six for 18 or "superior adult," and
sixteen alternative tests interspersed throughout the scale.
A person obtains his total score or "mental age" by taking
all the tests in perhaps four or five age level groups. For
instance, if a person passed all the tests at the 9 year level,
five out of the six at the 10 year level, four out of the eight
at the 12 year level, two out of the six at the 14 year level,
and failed all tests above 14, his total score or "mental age"
would be llj^. In assigning a given test to any age level,
all the tests were first tried out, and the positions of the
tests juggled about so that the ten year old children tested
had an average score of 10, the eleven year old children an
average score of 11, etc.

When we say that a person has a "mental age" of eight
on the Stanford-Binet scale, we do not mean that he has
the mentality of a child of eight, but that he made a total
score on that scale equal to that of the average eight year
child tested in the particular group on which the scale was
standardized. In all, about 1000 children, approximately
equally distributed in the chronological ages from 5 to 14,
formed the basis of the Stanford standardization. This
standardization is a very excellent method of measuring

56 AMERICAN INTELLIGENCE

intelligence, and worked very well with our army adult
group as shown by the score distribution of the 653 men
in the special experimental group given above, but we
should always regard the term "mental age" as a score,
not as a diagnosis.

By correlating the alpha test with the Stanford-Binet
scale, we can find the approximately equivalent score, or
"mental age" for each possible alpha score. The operation
resembles that of expressing values sterling in dollars. One
frequently hears the statement that the army tests proved
that the average citizen of this country has a mental age
equivalent to that of a child of thirteen. Nothing could be
more ridiculous. It is true that the average score of a sample
of 93,955 soldiers representing the entire white draft, when
translated into the Stanford-Binet scale, is 13.14. This
means that the approximately equivalent score on the
Stanford-Binet is 13.14. To say that the average citizen
has a mentality of the child of 13 is putting the cart before
the horse, for we are grading 93,955 people, and by infer-
ence the entire country, on a standard fixed by some 82
fourteen year old children who happened to be tested in
California.

In addition to the 1000 children on whom the Stanford-
Binet scale was standardized, the tests at the 16 year
"average adult" level, and the 18 year "superior adult"
level were standardized on 30 business men, 150 "migrat-
ing" unemployed men, 150 adolescent delinquents, and 50
high school students. It is thus seen that the Stanford-
Binet standardization rests on a number of cases too small
to upset the army standards based on 93,955 cases.

The term "mental age" is bad scientific slang for a total
score. Psychologists are gradually abandoning the age
standardization of tests. At the same time, publicists in
various fields, although novices in psychology, are drawing

AMERICAN INTELLIGENCE 57

rather vicious conclusions from "mental age" findings. It
is an unfortunate situation.

The methods used in creating and standardizing psycho-
logical tests are entirely empirical, and therefore rather
hard to explain to the layman, who is familiar only with
the "school teacher" type of examination. The school
teacher writes an examination, and lets it stand as an ab-
solute measure. The psychologist makes an examination,
tries it out, and judges each individual member of the
group as compared with the other members of the group.
As more and more people are examined his standards of
judgment become more reliable. In other words, his stand-
ards are those that he gets, not those that he thinks he
ought to get. Therefore, instead of deploring the fact that
the average person has a "mental age" of thirteen, we can
simply say that the conversion of the results of the army
test into the Stanford-Binet scale shows an average score
of 13, and that this is the score to be expected from the
average adult.

Another very common mis-statement prevalent con-
cerning the army results is that they proved that 24.9%
of the drafted men were illiterate. Among the men sent to
examination beta would be found, first, English speaking
illiterates, second, non-English speaking individuals, either
literate or illiterate in their own tongue, third, defectives,
and fourth, cases accidentally sent to the wrong examina-
tion. The method of selecting men for beta varied from
camp to camp, and sometimes from week to week in the
same camp. There was no established criterion of literacy,
and no uniform method of selecting illiterates. In a group
of 1,552,256 men examined, 386,196 or 24.9% were, for
some reason or other, sent to beta. The army definition of
literacy as ^'ability to read and understand newspapers and
write letters home'' can not be identified with the fact of having

58 AMERICAN INTELLIGENCE

been sent to beta. The statistics of the army examinations
give us no accurate figures on the percentage of ilKteracy.
The individual examination for ilHterates and non-
Enghsh speaking, the performance scale, was a composite
scale, the tests of which were drawn from workers in vari-
ous fields, particularly from H. A. Knox, who had worked
on non-verbal performance tests at Ellis Island, R. Pintner
and D. G. Paterson, who had developed a scale of perform-
ance tests, William Healy, H. H. Goddard, and other in-
vestigators. The performance examination was given some-
times as the long scale (8 or 10 tests) and sometimes as the
short scale (5 tests). The short scale showed a correlation
of 0.97 with the long scale, so that the reduction of the
number of tests to save time in examining was entirely jus-
tified. A short-cut method was also used in giving the
Stanford-Binet examination which was quite satisfactory
(correlation 0.91). The short performance scale showed a
correlation of 0.84 with the Stanford-Binet scale. The
Yerkes-Bridges point scale which was sometimes used in-
stead of the Stanford-Binet was also abbreviated, and the
abbreviated point scale showed a correlation of 0.934 with
the complete point scale. In general, the methods of indi-
vidual examining in use were quite reliable, and so closely
related to the Stanford-Binet scale that the results could
be converted into Stanford-Binet scores without any ap-
preciable source of error. In all calculations in this study,
scores from the point scale and the performance scale ex-
aminations have been treated by converting into Stanford-
Binet scores.

SECTION IV

RELIABILITY OF THE MEASURES

The reader, who has followed the discussion of the indi-
vidual tests through the preceding pages, will be convinced
that most of the tests used were satisfactory. In general,
the eight alpha tests, the Stanford-Binet scale, and tests
4, 5, 6 and 7 of beta gave complete or limited distributions
which approximated the Gaussian normal curve. Figure 24.

Figure 24. The normal distribution curve. A type of distribution
given by all alpha tests^ tests 4, 5, 6 and 7 of beta^ and the Stan-
ford-Binet scale.

Beta test 3 gave a distribution which could not be inter-
preted, while beta tests 1 and 2, being too easy, gave a
skewed distribution of the approximate form of the curve
shown in Figure 25.

It is not necessary here to enter into any lengthy dis-
cussion of the method of converting the results of these
sixteen tests into the combined scale. The reader interested
in the statistical methods used is referred to Chapter 2,
Part 3 of Memoir XV (pp. 573-657). If the reader is satis-

60 AMERICAN INTELLIGENCE

fied that 13 of the 16 tests give distributions conforming
in a general way to that shown in Figure 24, that is all that
is necessary. The tests can very obviously be equated, and
a combined scale constructed. By treating each of the eight
tests of alpha, each of the seven tests of beta, and the
Stanford-Binet test as different measuring scales, the com-
bined scale was evolved, based on the inter-relations of
these sixteen scales as shown by refined methods of cor-
relation.

Figure 25. A skewed distribution curve. A type of distTibution
given by beta tests 1 and 2.

The army results are reported in tables showing the
number of men scoring in certain class intervals, i. e. be-
tween and 4, 5 and 9, 10 and 14, etc., up to the interval
205 to 212 on alpha; between and 4, 5 and 9, 10 and 14,
etc., up to 115-118 on beta. In the same way, the scores in
other tests are reported in class intervals. The study of the
1047 cases showed how individuals falling in each of the
class intervals were distributed on the theoretical com-
bined scale, i. e., it was possible to find the combinations
of tests from which individuals in each class interval would

AMERICAN INTELLIGENCE 61

obtain their scores. The combined scale was therefore built
empirically on the results of the 1047 cases. Tables were
constructed on this basis showing how individuals falling
in each class interval of each of the three examinations
should be redistributed on the combined scale. It is then
possible to take a group which had been examined partly
by alpha, partly by beta, and partly by the Stanford-
Binet examination, and determine how that group would
have scored on the combined scale if all individuals in the
group had been given all three examinations. The com-
bined scale is the most accurate method available for treat-
ing the data derived from the army examinations.

In this study the data from the principal sample have
been re-figured on the combined scale by the method de-
scribed on page 65^ of Memoir XV:

"In each group the alpha distribution was distributed
on the combined scale by the use of table 159, the beta dis-
tributions by table 162, and the Stanford-Binet mental age
distribution by table 163. The performance scale distribu-
tions and the point scale distributions were handled in the
following way: the performance distributions were first
transformed into Stanford-Binet mental age distributions
by the use of the regression formula :

T.yr i 1 A / N 0.50 Performance score + 72
Mental Age (m years) =

This formula was derived from the correlation of a sample
of 350 cases who had both Stanford-Binet mental age rat-
ings and performance scale ratings. The point scale distri-
butions were transformed into Stanford-Binet mental age
distributions by the use of the table in the examiner's guide.
Part I, pages 195ff. These transformations only approxi-
mate the truth, but owing to the fact that the performance

62 AMERICAN INTELLIGENCE

and point scale cases constitute less than 3 per cent of any
group handled it would take a considerable error in trans-
formation seriously to affect the whole."

The conversion of the data of the principal sample into
the combined scale reported in Memoir XV contains some
inaccuracies. The statistical labor involved in the evolution
of the combined scale was so great that the method was not
available until the report was practically completed. The
calculations were made by different individuals working
under pressure, and errors were unavoidable. It has there-
fore been considered worth while to repeat these calcula-
tions at leisure, checking each operation carefully and
carrying the analysis of some of the groups of the principal
sample further than that reported in Memoir XV.

It is now necessary to review very briefly the results of
checking the army mental tests against outside criteria.
We might have a measuring scale, all elements of which
gave perfect score distributions, and which were highly
inter-correlated, but even then we would need outside cri-
teria to prove that we were measuring intelligence. Enough
material is already on hand to prove that the army tests
were reliable measures of intelligence. In the following dis-
cussion we will cite several instances.

The best proof of the validity of the test series comes
from a study of the relation between the inteUigence rat-
ings and education. The correlation of the combined scale
with reported school grade was 0.75 (based on 653 cases
from the special experimental group of 1047 men). The
correlation between alpha scores and schooling for this
group was 0.75, the eight tests of alpha separately com-
pared with schooling all showing correlations between 0.60
and 0.74. The correlation with beta total scores and school-
ing for this same group was 0.67, and that between Stanford-
Binet scores and schooling 0.65. These correlations show

AMERICAN INTELLIGENCE 63

a positive relationship between intelligence as measured by
the various methods and years of schooling.

Very few people realize the severity of the elimination
process that goes on from year to year in our schools and
colleges. The study of the schooling of the native born white
draft, as sampled by upward of 80,000 cases, showed the
following startling facts: of every thousand native born
recruits who entered the first grade, 970 remained in school
till grade two, 940 till grade three, 905 till grade four,
830 till grade five, 735 till grade six, 630 till grade seven,
and 490 till grade eight; 230 of them entered high school,
170 kept on till the end of the second year, 120 till the end
of the third year, and 95 of the original thousand graduated
from high school; 50 of these entered college, 40 kept on
till the end of the second year of college, 20 till the end of
the third year, and 10 graduated. It is, of course, impossible
to determine how many of those that leave school leave on
account of lack of pecuniary opportunity, or on account
of lack of intelligence. It is ridiculous to assume that 1000
men in 1000 have sufficient intelligence to graduate from
college, and equally absurd to assume that only 10 in 1000
have such a high intellectual endowment that they can
graduate from college. But, inasmuch as we can not deny
the intellectual elimination, we must expect a very high
correlation between intelligence and schooling.

The army tests uniformly show officers superior to en-
listed men. This is to be expected, for officers were selected
for ability. Nevertheless, one may perhaps contend that
the high scores of the officers were due to superior educa-
tion and not to greater intelligence. Very nearly half of
the officers were college graduates, and another quarter had
begun but not completed a college course. The objection
that the superior scores of officers were due to education
rather than intelligence may be effectively answered by a

64 AMERICAN INTELLIGENCE

crucial test, which was made by the army investigators
when they compared the alpha scores of 660 officers who
had never gone beyond the eighth grade in school with the
alpha scores of 13,943 native born recruits all of whom had
gone beyond the eighth grade. The results of this comparison
are reported (p. 779 of Memoir XV) as follows: "Every
recruit in the recruit group has had more schooling than
any officer in the officer group; the least educated recruit
in the group has had a longer education than the best edu-
cated officer included. And the group of officers neverthe-
less makes a slightly higher record on examination alpha.
It is evident then that the examination is measuring other
qualities, in which officers stand above recruits, to a
greater extent than it is measuring education." The distri-
butions of the alpha scores of these two groups are shown
in Figure 26. In general, the comparison of the army test
scores with education indicates that the tests are genuine
measures of intelligence.

The army investigators were, of course, called upon early
in the war to prove that the tests they recommended were
genuine tests of intelligence. For the assistance of army
examiners and administrative officers having before them
the problem of the assignment of men, a small pamphlet.
Army Mental Tests (Washington, D. C, 1918, pp. 24)
was prepared, presenting in graphic form the results of
several different methods used in some of the camps for
establishing the reliability of the tests by checking them
against outside criteria. In the following pages, some of the
charts from this booklet have been reproduced, and the
method of interpreting the charts is described briefly. All
of the methods reported use the letter grade classffication
(A, B, C, etc.), which is less accurate than the combined
scale method used in this study, but they tend to show in

OFFICERS

WHITE DRAFT. NATIVE BORN

Figure 26. Examination alpha as independent of education. Com-
parison of alpha scores of officers of eighth grade schooling or
less with alpha scores of native born white recruits of ninth
grade schooling or more. "Although these groups overlap in
schooling not at all, the officers make nevertheless slightly higher
scores on alpha." (Quotation from p. 779, and figure from p.
778, Memoir XV.)

66 AMERICAN INTELLIGENCE

a general way the trend of the results, and that is all that
is necessary in this ease.

In the long run, we should expect a small positive cor-
relation between intelligence and rank. Intelligence is by
no means the sole determiner of military success, but since
it is one element in the complex of abilities required, we
would expect to find a general tendency toward high scores
with higher ranks. Figure 27, which is reproduced from the

ENLISTED MEN (18792) Relatively Illiterate
ENLISTED MEN (82936) Litebate
CORPORALS (4023)

SERGEANTS (8393)

O. T. C. (9240)

OFFICERS (8819)

Figure 27. Distribution of intelligence scores according to rank.
The officers are above the candidates in the Officers' Training'
Camps (O. T. C), the candidate officers are above the sergeants,
the sergeants above the corporals, and the corporals above the
enlisted men. (From p. 8 of pamphlet. Army Mental Tests.)

booklet referred to, shows the distribution of scores of vari-
ous ranks on the rough A, B, C scale. The oflScers form a
group quite distinct from the general run of enlisted men,
and they are also above the candidates for commissions in

AMERICAN INTELLIGENCE

the OflScers' Training Camp (O. T. C.) group. The sergeants
are above the corporals, and the corporals above the en-
listed men.

The Officers' Training Camps give an additional check
on the intelligence tests. In the schools examined, the can-
didates were recommended for a period of special training
for commissions by the regimental organizations. The selec-
tion of the candidates was very rigid, then, in the first in-
stance. Figure 28 shows roughly the results of applying the

A B C+ C C D

PERCENT
SUCCESS

PERCENT
FAILURE

O. T. C.
1375 MEN

Figure 28. Success in Officers' Training Camps as predicted by
examination alpha. Each vertical bar represents all (100%) of
the candidates who tested A, B, C-|-, etc. All men above thehori-
zontal line eventually received commissions, and all men below
failed. 91^% of the men above C-\- received commissions.
58^^ of the men below C-j- failed. (From p. 9 of pamphlet,
Army Mental Tests.)

68 AMERICAN INTELLIGENCE

army tests to the training camp groups. Each solid vertical
bar represents all the men of a given letter grade. There
were, of course, more A and B men than C and D men,
but for purposes of comparison each letter group is treated
as a whole (100%). All men above the horizontal line re-
ceived commissions at the close of the Officers' Training
Camp, and all the men below the line failed to receive com-
missions. Figure 28 shows clearly that about nine out of
ten A and B men eventually received commissions, while
for C and D men the chances were very slight.

Figure 29 shows in a general way the manner in which
groups selected by various outside criteria contributed to
the upper end of the intelligence scale (A and B), and to
the lower end of the scale (C , D and E). We are already
familiar with the differences between ranks shown in this
figure. Sixty company commanders were asked to designate
their ten "best" and ten "poorest" privates. The results of
this comparison of the ten "best" and ten "poorest" pri-
vates also appears in Figure 29. The other two classifica-
tions, "men of low military value," and "unteachable men"
represent a type of officers' rating. In general the test re-
sults check with officers' ratings independently made.

The results presented, showing the relation between rank
and intelligence, and officers' ratings and intelligence, indi-
cate clearly a certain positive relationship between tests
and military success. Recognizing the fact that intelligence
is only one factor tending to produce military success, we
accept the results of checking the tests against military cri-
teria as additional proof that the tests are genuine measures
of intelligence.

A rough but rather interesting check of the army in-
telligence tests may be made by glancing at the scores of
men classified by occupations. Figure 30 gives the range of
the intelligence scores of the middle 50% of various occu-

D.D . E C+,C, C

^ AND B

COMMISSIONED OFFICERS

8819
I 1

O. T. S. STUDENTS
9240

SERGEANTS
3393

CORPORALS
4093

TEN BEST PRIVATES

606

WHITE RECRUITS

77299

DISCIPLINARY CASES
491 Camp Bix

TEN POOREST" PRIVATES

582

I I e

MEN OF LOW MILITARY VALUE'*^

*147 Camp Custer
t I

UNTEACHABLE MEN

255 Camp Hancock

rzZD

Figure 29. Comparison of army tests records with various inde-
pendent criteria. The distributions of scores by ranks are shown
in another way in Figure 27. The men were rated as "ten best,"
"ten poorest," "of low military value," and "unteachable" by
their officers. The chart shows a close correspondence between
the brief psychological examinations and officers' judgments
made after weeks of observation. (From p. 10 of pamphlet,
Army Mental Tests.)

> - I

I c. I

c-<

Laborer . .
Gen. miner
Teamster .
Barbe* ...

C <

Horseshoer .....
Bricklayer

Cook

Baker

Painter

Gen. blacksmith .

Gen. carpenter . .
Butcher

Gen. machinist

Hand riveter

Tel. and tel. lineman , . .

Gen. pipefitter

Plumber

Tool and gauge maker.

Gunsmith

Gen. mechanic

Gen. auto repairman .
Auto engine mechanic

Auto assembler

Ship carpenter

Telephone operator . .

C <

Concrete const, foreman
Stock-keeper

Photographer

Telegrapher

R.R. clerk

Filing clerk

Gen. clerk

Army nurse

Bookkeeper

B <

Dental officer

Mechanical draftsman

Accountant

Civil engineer

Medical officer , ,

A ^ Engineer officer

fD^T

1 cT-r

Figure 30. Success in civil occupations compared with army test
records. The figure shows in a general way the correlation be-
tween intelligence as measured by the army tests and intelli-
gence as indicated by position in civil life. (From p. 829,
Memoir XV.)

AMERICAN INTELLIGENCE 71

pational groups, the position of the man half way up or
down the scale being marked by a short vertical line. In
some of the occupational groups, the number of cases is
small, and the classification itself may be at fault in many
instances, but the chart nevertheless shows a general ten-
dency of the sort we should expect to find, for a process of
intellectual selection occurs in industry which is just as-
rigid as that occurring in our public schools.

We have briefly inspected the different sorts of evidence
from independent fields which indicate that the army tests
were genuine measures of intelligence. Further discussion
of this point is unnecessary. The army tests were not in-
fallible, and mistakes in classifying men were undoubtedly
made, but the tests were satisfactory rough measures.
When used in comparing groups as the tests are in this
study, their reliability is increased, for errors in measure-
ment would tend to equalize in each group. We should ex-
pect the same percentage of error in classifying recruits
born in Russia as we should recruits born in Sweden. Thus
we use the tests as general measures of group tendencies,
and as group measurements the tests have a suflSciently
high degree of reliability to make positive conclusions
possible.

In the foregoing pages the army tests have been de-
scribed briefly, the method of treating the results from vari-
ous examinations by the combined scale reviewed, certain
misconceptions discussed, and a few bits of supporting
evidence assembled. Persons interested in a further study
of the tests should consult a little book by Yoakum and
Yerkesi, or. Memoir XV. We may now proceed to analyze
American intelligence by treating the psychological exam-
inations made in the army as a mental census of the
population of this country.

IC. S. Yoakum and R. M. Yerkes. Army Mental Tests. New York, 1920, Pp.203.

PART II

STATISTICAL ANALYSIS OF THE
ARMY TEST RESULTS

PART 11. SECTION I
THE PRINCIPAL SAMPLE

All results from the psychological examinations in the
camps were sent to Washington. It was impracticable as
well as undesirable to tabulate the results in the case of
every man examined. An intelligent selection or sampling
of cases will give results more nearly typical of the country
at large, than the entire group tested, which would be un-
duly weighted for the more populous States, for camps
giving the greatest number of examinations, particular
draft quotas, etc. In order to obtain a sample for the white
draft and the negro draft, cases were "randomly" (or better,
impartially) selected in accordance with certain definite
principles. The groups were as follows:

Group I: White draft, pro-rated, by States. . . . 41,278
Selected from 15 National Army
camps, according to the State from
which drafted, and according to the
ratio of one recruit per thousand
white male population.

Group II: White draft, additional, by States. . . 14,684
Additional selection of cases in-
tended to bring the total represen-
tation from each State up to 1,000
cases.

Group III: White draft, additional, by camps. . . 40,392
Additional selection of cases in-
tended to bring the entire samp-
ling of the white draft up to ap-
proximately 100,000 cases.

76 AMERICAN INTELLIGENCE

Group IV: Negro draft, pro-rated, by States. . . . 19,992
Selected in the same manner as
Group I.
Group V: Negro draft, additional, for Northern

States 5,400

Chosen to represent the negro draft
of the north.
Other groups were selected to meet other problems as
follows :

Group VI: White officers 15,528

Selected proportionately to their
occurrence in different arms of the
service, with some additions to sup-
plement the smaller arms, and the
Medical Department.

Group VIII: White established organizations 24,205

Selected to provide comparison be-
tween enlisted men of various arms
of the service.

Group X: Special experimental Group 1,047

Randomly selected individuals of
the white draft born in English
speaking countries, who were given
both alpha and beta, and, where
possible, the Stanford-Binet ex-
amination.
These groups selected as representative of the country
at large were analyzed by the Hollerith system of mechan-
ical sorting.

In this study we are concerned with:
Groups I, II and III representing the white draft.
Groups IV and V representing the negro draft,
Group VI representing the white officers, and Group X,
the special experimental group.

SECTION II

ANALYSIS OF THE MAIN GROUPS OF THE
PRINCIPAL SAMPLE

The tabulations in Memoir XV showing the distribution
of scores on each type of examination of the white draft
(Groups I, II, and III), the negro draft (Groups IV and
V), and the white officers (Group VI) have been re-figured
on the combined scale. The following tables were used:

For the white draft:

Alpha: Table 183 (p. 666) for men who took

alpha only 67,254

Beta: Table 184 (p. 666) for men who took beta

only, or alpha and beta 23,547

Individual:

Table 185 (p. 667) for men who took Stanford-
Binet examination only, or follow-
ing alpha, following beta, or follow-
ing alpha and beta 1,246

Table 186 (p. 667) for men who took point
scale examination only, or following
alpha, following beta, or following
alpha and beta 689

Table 187 (p. 668) for men who took perform-
ance scale examination only, or fol-
lowing alpha, following beta, or fol-
lowing alpha and beta 1,237

Total Cases 93,973

78 AMERICAN INTELLIGENCE

For the negro draft:

Alpha: Table 239 (p. 716) for men who took

alpha only 8,429

Beta: Table 241 (p. 717) for men who took beta

only, or alpha and beta 14,350

Individual :

Table 242 (p. 717) for men who took Stanford-
Binet examination only, or follow-
ing alpha, following beta, or follow-
ing alpha and beta 403

Table 229 (p. 711) for men who took point
scale examination only, or following
alpha, following beta, or following

alpha and beta 390

Table 228 (p. 710) for men who took perform-
ance scale examination only, or fol-
lowing alpha, following beta, or fol-
lowing alpha and beta 32

Total Cases 23,604

For White officers :

Alpha: Table 182 (p. 665) for all officers who

took examination alpha only. 15,544

It will be remembered that examination beta was given
to all men who had been selected as illiterate or non-Eng-
lish speaking before examination alpha was given, and also
to those who took alpha and failed to make a reliable score.
In the same way individual examinations were given to the
lowest scoring cases in beta. Consequently, in figuring the
results, if a man has taken both alpha and beta, his alpha
score is disregarded and his beta score taken. In the same
way, if he took both alpha and beta, and was then given an

AMERICAN INTELLIGENCE 79

individual examination, his alpha and beta scores are dis-
regarded, and his score on the individual examination is
taken as expressing the best measure of his intelligence. In
other words we use alpha as our measure if alpha only was
given, beta as our measure if beta alone, or alpha and beta
were given, and the individual examination as our measure
if it was given at all, on the assumption that the most re-
liable test of a man was the last one given.

The distribution of the scores of the white officers, white
draft and negro draft is shown in Table 1. The first three
columns are read in this manner: six officers and one re-
cruit measured between 24.0 and 24.9 on the combined
scale, one hundred and six officers and eighteen recruits be-
tween 23.0 and 23.9, etc. The first three columns show the
actual distributions, i. e. six officers out of 15,543, one re-
cruit out of 93,955, etc. The last three columns show each
of these distributions arranged according to the number in
each ten thousand who scored at each class interval of the
combined scale. The last three columns read in this man-
ner: four officers in ten thousand test between 24.0 and
24.9 on the combined scale; sixty-eight officers and two re-
cruits in ten thousand score between 23.0 and 23.9 on the
combined scale, etc. The "proportion in each ten thousand"
may be read as a percentage by pointing off two decimal
places.

Table 1 also shows the average score of the white officers,
white draft and colored draft on the combined scale to be
18.84, 13.54 and 10.41 respectively. The standard devia-
tion (S. D.) is also shown. An average has little significance
without reference to a measure of variability of the series
of measurements on which it is based. The conventionally
accepted measure of variability is the standard deviation,
which is derived by taking the square root of the average
of the squares of the individual deviations from the aver-

Table No. 1

Distribution of the intelligence scores of the main groups
of the principal sample on the combined scale.

PROPORTION IN EACH

ACTUAL

DISTRIBUTION 1

COMBINED

TEN THOUSAND

SCALE

INTERVALS

WHITE

NEGRO

WHITE

NEGRO

OFFICERS

DRAFT

OFFICERS

i DRAFT

DRAFT

24.0-24.9

23.0-23.9

106

....
....

\ "k

22.0-22.9

612

124

S94

t 13

....

21.0-21.9

1648

444

106C

) 48

20.0-20.9

2522

1006

wrc

J 107

19.0-19.9

2836

1804

1824

t 192

18.0-18.9

2698

2996

i73e

J 319

17.0-17.9

2155

4687

172

138';

^ 499

16.0-16.9

1454

6847

330

93^

i 729

140

15.0-15.9

837

9328

600

538

} 993

254

14.0-14.9

412

12019

1031

26^

) 1279

437

13.0-13.9

179

14659

1793

11^

) 1560

760

12.0-12.9

14002

2572

) 1490

1090

11.0-11.9

9481

2951

) 1009

1251

10.0-10.9

6227

3187

I 662

1351

9.0-9.9

4433

3319

L 472

1406

8.0-8.9

....

2876

2891

306

1225

7.0-7.9

....

1683

2149

179

911

6.0-6.9

814

1315

557

5.0-5.9

....

334

684

290

4.0-4.9

....

122

302

128

3.0-3.9

....

112

2.0-2.9

....

1.0-1.9

No. cases

15543

93955

23596

Average

18.84

13.54

10.41

S.D

2.10

2.92

2.79

.^ g o fl

82 AMERICAN INTELLIGENCE

age. The size of the standard deviation indicates the vari-
abihty of the group. In this case the white officers are more
homogeneous (less variable) than the white draft and the
negro draft (standard deviation of 2.10 compared with 2.92
and 2.79). Furthermore, the negro draft is more homogen-
eous than the white draft (2.79 compared with 2.92).

v^^ Figure 31 shows graphically the proportions given in the
last columns of Table 1. The horizontal line shows the com-
bined scale intervals and the vertical lines the number in
each ten thousand. These curves show very clearly the dis-
tribution of intelligence in the three groups. In general, the
distributions are similar in shape, but they differ markedly
in their position on the scale of intelligence. The differences
are very great. Of the officers, 98.87% are above the average
of the white draft, and 99.97% are above the average of the
negro draft. Of the white draft, 86.31% are above the aver-
age of the negro draft. Only 13.13% of the negro draft are
above the average of the white draft. This method of figur-
ing gives us some indication of the differences between the
groups. If the distribution of intelligence in two groups were
the same, 50% of either group would exceed the average
of the other group. If the distributions were absolutely dis-
tinct, and there was no over-lapping, then 100% of one
group would exceed the highest man in the other group.
This last case would only occur if we compared very ex-
treme groups (such as officers and idiots), and the conven-
tional method is that of expressing the difference on a scale
of 50%, i. e. the per cent, of one group above the average

^ of the other group.

^ The most reliable method of determining the relation-
ship between two groups is that of comparing the differ-
ence between the averages with the probable error of the
difference. This method takes account of the variability of
the original measures in each series, and also the reliability of

AMERICAN INTELLIGENCE 83

the measures as determined by both the variability and the
number of cases. If the difference is 2 and the probable
error of the difference is 1, the difference would be written
2 1, meaning that the chances are even that it would
not be less than 1, or more than 3. Difl^erences which
are not at least four times as great as the probable error of
the difference are not conventionally accepted as significant.

Applying this method to the groups under consideration, /^-~
we find the following differences :
Between white officers and white draft 5.30 .0131.

(The difference is 405 times the probable error of the

difference.)
Between white officers and negro draft 8.43 .0167.

(The difference is 505 times the probable error of the

difference.)
Between white draft and negro draft 3.13 .0138.

(The difference is %%1 times the probable error of the

difference.)

SECTION III

ANALYSIS OF THE WHITE DRAFT INTO
FOREIGN AND NATIVE BORN

The next problem is that of breaking up the white draft
into its constituent elements as far as possible. In Chapter
6, Part 3 of Memoir XV we find tables showing the dis-
tribution of the scores of 12,492 recruits who reported that
they were born in foreign countries. The tables in Memoir
XV give the scores of all but 800 who reported foreign
birth. As all of the 12,492 reported cases were in Groups
I, II and III, they may be deducted from those groups,
leaving a remainder composed of (1) native born, and (2)
foreign born who failed to report on their psychological ex-
amination blanks the fact that they were born in some
country other than the United States. How large this latter
group is we have no way of estimating. In the following
tabulations the term "native born" is defined as all who
stated that they were born in the United States, plus all
who failed to record the country of their birth.

The original data giving the score distributions of the
12,492 foreign born were obtained from the following tables:

Alpha: Table 207 (p. 693) for men who took

alpha only 4,191

Beta: Table 208 (p. 694) for men who took beta

only, or alpha and beta 7,264

Individual:

Table 209 (p. 694) for men who took Stanford-
Binet examination only, or follow-
ing alpha, or following beta, or fol-
lowing alpha and beta 207

AMERICAN INTELLIGENCE 85

Table 210 (p. 596) for men who took perform-
ance scale examination only, or fol-
lowing alpha, following beta, or fol-
lowing alpha and beta 830

Total foreign born white draft 12,492

The actual distributions of the scores of the native born
and foreign born are shown in the first two columns of
Table 2, while the last two columns in this table give the
proportion in ten thousand scoring at each class interval.
These distributions are also shown graphically in Figure 32.
Here again it is apparent that we have two groups that are
markedly different. Of the native born 74.8% exceed the
average of the foreign born. The difference between the
native born and the foreign born is 1.72 .0186, a differ-
ence that is 92^/^ times the probable error of the difference.

Comparing the native born white draft with the negro
draft shown in the preceding section, we find that 88.76%
of the native born white draft exceed the average of the
negro draft. The difference between these two groups is

3.36 ==.014, a difference that is 240 times the probable
error of the difference.

Comparing the foreign born with the negro draft, we find-^^
that 70.44% of the foreign born exceed the average of the
negro draft. The difference between these two groups is

1.64 .0212, a difference that is 77 times the probable
error of the difference.

Table No. 2

Analysis of the white draft into foreign born and native
born groups. Distribution of each group on the combined
scale.

PROPORTION IN EACH

ACTUAL DISTRIBUTION 1

TEN THOUSAND

COMBINED

SCALE

INTERVALS

NATIVE

FOREIGN

NATIVE

FOREIGN

BORN

WHITE

DRAFT

24.0-24.9

.....

23.0-23.9

22.0-22.9

120

21.0-21.9

428

20.0-20.9

971

119

19.0-19.9

1733

213

18.0-18.9

2850

147

349

117

17.0-17.9

4403

284

540

227

16.0-16.9

6345

502

779

402

15.0-15.9

8537

791

1048

633

14.0-14.9

10870

1148

1334

919

13.0-13.9

13066

1593

1604

1275

12.0-12.9

12220

1782

1500

1426

11.0-11.9

7885

1596

968

1277

10.0-10.9

4801

1425

589

1141

9.0-9.9

3178

1254

390

1004

8.0-8.9

1985

892

244

714

7.0-7.9

1153

530

142

424

6.0-6.9

556

259

207

5.0-5.9

228

106

4.0-4.9

3.0-3.9

2.0-2.9

1.0-1.9

No. of cases. . . .

81465

12492

Average

13.77

12.05

S. D

2.86

2.88

j9?

/ /

^ o

o ^

^ ^ .S ^ i 3

0) >.

a; ^ ?^ fl, ^ ^

<i; .-S

i %

rl3 o
G CO

44H O

.2 Id

f3 en

GO 4)

cnH ^ I "^ ^

^ '^ ^ I a,t

H ^ +S g 53 03

'-' ^ >^

o o

OJ W5 4J

<U o -4^

^ be

S a; ^

fl.5 fl

^ ^

=^ be tn rrt - ^

b d fl (u be ;h 2
g g^'S 3 'S .2 "

O i-i n =d - '^

b* ^

Pi Ph O H 4-l 4H

o .<u ^

SECTION IV

ANALYSIS OF THE FOREIGN BORN WHITE

DRAFT INTO YEARS OF RESIDENCE

GROUPS

The next problem is that of breaking up the foreign born
white draft into its sub-groups, in order to discover the rea-
son for the discrepancy between the main group of foreign
born and the native born white draft. In Chapter 7, Part 3
of Memoir XV we find tables showing the scores of 11,295
foreign born (included in Groups I, II and III of the prin-
cipal sample) classified according to the number of years of
residence. The scores are tabulated for the following groups:

1st : a six year period to 5 years of residence

2nd: a five year period .... 6 to 10 years of residence
Srd: a five year period .... 11 to 15 years of residence
4th: a five year period .... 16 to 20 years of residence
5th: a ten year period. . . .Over 20 years of residence

As the age limit of the draft was 31, the last group would
include cases who have been in the United States since
childhood. It is probable that all of the 11,295 cases are in-
cluded in the 12,492 cases shown in the preceding section.
The discrepancy in numbers between the two groups is
probably due to the fact that some foreign born reported
the country of their birth, but failed to report the number
of years they had been residents of the United States.

The scores reported in Chapter 7 have been re-figured on
the combined scale. The following tables were used for the
original data:

AMERICAN INTELLIGENCE 89

Alpha :Table 219 (p. 701) for men who took alpha

only 3,619

Beta: Table 220 (p. 702) for men who took beta

only, or alpha and beta 7,264

Individual :

Table 221 (p. 702) for men who took performance
scale examination only, or following
alpha, following beta, or following alpha
and beta 802

Table 222 (p. 703) for men who took Stanford-
Binet examination only, or following
alpha, or following beta, or following
alpha and beta 207

Total cases in all residence groups 11,295

The distributions of the scores of the five years of resi-
dence groups on the combined scale are shown in Table 3.
This table shows a very remarkable fact, viz., a steady
increase in the average scores with increasing years of
residence, the averages being:

TO 5 YBS. 6 TO 10 YRS. 11 TO 15 YRS. 16 TO 20 YES. OVER 20 YES.

11.41 (2.85) 11.74 (2.80) 12.47 (2.77) 13.55 (2.60) 13.82 (2.71)

From to 20 years of residence, the average rises steadily
and the variability becomes less and less.

Table 4 gives the difference between each group and
every other group, together with the probable error of the
difference, and the ratio of the difference to the probable
error of the difference. All of the differences shown in Table
4 are significant except the difference between groups "16
to 20 yrs." and "over 20 yrs.," this difference (0.27) being
only three times the probable error of the difference
(0.0915).

Table No. 3

Analysis of the foreign born white draft by years of resi-
dence in the United States. Distribution of each resi-
dence group on the combined scale.

TEARS

OF RESIDENCE

COMBINED

SCALE

INTEKVALS

TO 5

6 TO 10

11 TO 15

16 TO 20

OVER 20

YEARS

4.0-24.9

0.1

23.0-23.9

....

o'.i

0.5

....

22.0-22.9

0.6

0.8

0.5

21.0-21.9

2.6

2.9

2.6

1.6

3.6

20.0-20.9

5.8

8.2

6.3

3.6

7.5

19.0-19.9

14.4

18.6

13.1

8.3

12.1

18.0-18.9

27.6

37.6

27.9

18.0

24.7

17.0-17.9

55.5

72.0

50.6

36.7

41.1

16.0-16.9

104.5

141.8

83.1

61.2

66.2

15.0-15.9

172.4

240.7

131.3

87.3

86.1

14.0-14.9

261.2

355.2

198.7

106.1

106.3

13.0-13.9

374.4

506.2

282.0

128.6

12.0-12.9

444.8

597.0

309.1

124.5

113.7

11.0-11.9

457.5

572.1

245.6

82.6

69.0

10.0-10.9

471.0

533.0

189.8

47.4

42.9

9.0-9.9

453.3

479.2

152.3

29.5

8.0-8.9

342.6

347.2

102.1

18.1

18.2

7.0-7.9

212.9

209.6

58.5

9.9

9.2

6.0-6.9

106.8

102.4

27.0

4.5

3.4

5.0-5.9

44.7

41.7

10.5

1.7

1.0

4.0-4.9

16.5

14.9

3.6

0.6

0.3

3.0-3.9

5.0

4.4

1.1

0.2

0.1

2.0-2.9

1.5

1.3

0.3

0.1

1.0-1.9

0.4

0.3

0.1

No. of cases

3576

4287

1897

771

764

Average

11.41

11.74

12.47

13.55

13.82

S. D

2.85

2.80

2.77

2.60

2.71

Table No. 4

Comparison of the average scores on the combined scale
of the five years of residence groups of the foreign born
white draft.

YEARS OF RESIDENCE
GROUPS

TO 5 YRS.
AVE. 11.41

3576 CASES

6 TO 10 YRS.
AVE. 11.74

4287 CASES

11 TO 15 YRS.

AVE. 12.47
1897 CASES

16 TO 20 YRS.

AVE. 13.55

771 CASES

6 to 10 yrs. (Diff.
Ave. 11.74 (P. E. Diff.
4287 cases (Ratio

+0.33
0.0431
7.7

11 to 15 yrs. (Diff.
Ave. 12.47 (P. E. Diff.
1897 cases (Ratio

+1.06
0.0536
19.8

+0.73
0.0517
14.1

16 to 20 yrs. (Diff.
Ave. 13.55 (P. E. Diff.
771 cases (Ratio

+2.14
0.0709
30.2

+1.81
0.0695
26.0

+1.08
0.0764
14.1

Over 20 yrs. (Diff.
Ave. 13.82 (P. E. Diff.
764 cases (Ratio

+2.41
0.0735
32.8

+2.08
0.0721
28.8

+1.35
0.0788
17.1

+0.27
0.0915
3.0

Table No. 5

Comparison of the average scores on the combined scale
of the native born white draft with the five years of resi-
dence groups of the foreign born white draft.

NATIVE BORN WHITE DRAFT. 81465

CASES, AVE.

13.77

RATIO OF

PROBABLE

DIFFERENCE

YEARS OF RESIDENCE
GROUPS

DIFFERENCE

ERROR OF
THE DIF-

TO PROBABLE
ERROR OF

FERENCE

THE DIF-
FERENCE

to 5 yrs.

Ave. 11.41

-2.36

0.0104

226.9

3576 cases

6 to 10 yrs.

Ave. 11.74

-2.03

0.0296

68.6

4287 cases

11 to 15 yrs.

Ave. 12.47

-1.30

0.0434

30.0

1897 cases

16 to 20 yrs.

Ave. 13.55

-0.22

0.0636

3.5

771 cases

Over 20 yrs.

Ave. 13.82

+0.05

0.0664

0.75

764 cases

AMERICAN INTELLIGENCE 93

Table 5 shows the relationship of the five years of resi-
dence groups to the native born white draft. With increase
in the time of residence, the differences between the native
born and the foreign born become increasingly less signifi-
cant. The difference between the native born and the "16
to 20 yrs." group of foreign born is slight (0.22) and is less
than four times the probable error of the difference
(0.0636). The foreign born group in this country over 20
years have an average score identical with the average
score of the native born, the actual difference (0.05) being
smaller than the probable error of the difference ( == 0.0664) .

Figure 33 shows graphically the relationships between
the averages of the years of residence groups and the native
draft. In this graph, the horizontal line represents in-
creasing length of residence and the vertical line repre-
sents increase in the average score on the combined scale.

This very remarkable fact of increase in the intelligence
score with years of residence was commented on by the
army authors in Memoir XV as follows :

"It is not possible to state whether the difference is
caused by the better adaptation of the more thoroughly
Americanized group to the situation of the examination or
whether some other factor is operative. It might be, for in-
stance, that the more intelligent immigrants succeed and
therefore remain in this country, but this suggestion is
weakened by the fact that so many successful immigrants
do return to Europe. At best we can but leave for future
decision the question as to whether the differences repre-
sent a real difference of intelligence or an artifact of the
method of examination." (p. 704.)

If our results reflect another factor independent of in-
telligence, which might be designated "the better adapta-
tion of the more thoroughly Americanized group to the
situation of the examination," we have no means of con-

NATIVE BORN WHITE DRAFT

.4^

^^^

0TO5

6 TO 10 n TO 15 16 TO 20

YEARS RESIDENCE IN UNITED STATES

OVER 20

Figure 33. Apparently increasing average intelligence with in-
creasing years of residence. The horizontal scale reads from
left to right according to increasing length of residence. The
vertical scale represents score on the combined scale. For
purposes of comparison, the position of the native born white
draft on the combined scale is shown by a continuous line. The
group of immigrants who have been in this country from 16 to
20 years have an average intelligence almost as high as that of
the native born^ while immigrants in this country over 20 years
test the same as native born.

96 AMERICAN INTELLIGENCE

trolling this factor. Ultimately, the validity of our conclu-
sions from this study rests on the validity of alpha, beta,
and the individual examinations. It is sometimes stated
that the examining methods stressed too much the hurry-
up attitude frequently called typically American. The ad-
justment to test conditions is a part of the intelligence test.
We have, of course, no other measure of adjustment aside
from the total score on the examinations given. If the tests
used included some mysterious type of situation that was
"typically American," we are indeed fortunate, for this is
America, and the purpose of our inquiry is that of obtain-
ing a measure of the character of our immigration. Inability
to respond to a "typically American" situation is obviously
an undesirable trait.

For our purposes then we will accept the definition of in-
telligence given on page 573 of Memoir XV, viz., "by 'in-
telligence' we mean the ability that manifests itself quanti-
tatively in a set of consistent scores in all of the types of
examination upon which our data are based." We are
forced to include the "adjustment to test conditions" in
our definition of intelligence. And we hope, probably in the
teeth of the facts, that the adjustment to test conditions
involved a situation that was "typically American."

The hypothesis that the more intelligent immigrants re-
main in this country while the more stupid ones go home,
which was offered by the army authors to account for the
increase of intelligence scores with increasing years of resi-
dence, can not be checked from the data available in this
study, and the emigration statistics give us little help.
Table 6 shows the ratio between emigrant aliens and im-
migrant aliens from each country from 1908 to 1917. No
figures for emigrant aliens are available prior to 1908.

Table 6 shows that since 1908 a very considerable num-
ber (approximately one third) of immigrants have event-

AMERICAN INTELLIGENCE 97

ually returned to their native countries. If the selection factor
mentioned were operating, and in the long run the depart-
ing group contained more persons of lower intellectual capac-
ity than the remaining group, an intelligence measurement
would show an increase in the direction shown in Table 3,
but the departing third would have to be very heavily
weighted with low grade individuals to make any consider-
able difference in the distribution of the remaining two
thirds.

Under the conditions of this study then, the hypothesis
that the more intelligent immigrants succeed and therefore
remain in this country must remain in the realm of specu-
lation, as it can not be demonstrated as a fact. For our
purposes the converse hypothesis, that the successful im-
migrants save their earnings and return to Europe to live
in comfort for the rest of their lives, is equally cogent.
Either hypothesis is a legitimate speculation. We must
agree with the army authors that the first hypothesis is
weakened by the fact that so many successful immigrants
do return to Europe. The first hypothesis is also weakened
by the fact that if the more inferior individuals left, the
distribution of the intelligence of the departing third would
have to be very markedly skewed at the lower end of the
scale. The distribution curve of the departing third would
be skewed to such an extent that only 10% or 15% of this
group would exceed the median of the remaining two
thirds. We must conclude that the selection factor mention-
ed might produce a slight change in the direction noted, but
that it is highly improbable that any such factor could pro-
duce a change in the amount observed.

The important problem which we are facing is that of
interpreting the fact of increase of intelligence test scores
with increasing years of residence. Does our curve in Fig-
ure 33 represent the growth of intelligence with increasing

Table No. 6

Per cent, that emigration was of immigration for fifteen
countries since 1908. The figures were obtained by di-
viding the number of emigrant ahens departed by the
number of immigrant ahens admitted. The result is the
ratio of emigration to immigration. The ratio 100 would
mean that the number of alien immigrants admitted
equalled the number of alien emigrants departed.

AUSTRIA

BELGIUM

CANADA

DENMARK

ENGLAND

GERMANY

GREECE

1908.

78.2
29.1

20.5
11.7

6.8

58.6

13.9
10.5

10.6
9.4

21.0
19.2

28.5

1909.

40.0

1910.

18.3

12.1

60.4

22.5

9.7

19.9

31.4

1911.

54.2

17.8

86.9

6.2

10.3

18.8

35.7

1912.

49.5

26.5

60.0

10.7

16.6

20.8

54.4

6Yr.

Period

42.8

17.5

57.9

12.8

11.3

19.9

37.2

1913.

23.0

10.8

63.6

9.4

13.7

13.8

134.0

1914.

27.0

19.9

37.0

10.0

20.2

14.4

31.0

1915.

74.7

13.9

28.3

12.4

35.8

18.2

77.8

1916.

16.1

2.4

15.5

15.4

39.8

15.3

17.8

1917.

18.9

3.8

18.1

17.8

33.5

16.9

8.5

5Yr.

Period

26.7

13.7

30.4

12.0

23.6

14.6

47.7

10 Yr
1908-

. Period
1917

36.8

15.9

40.3

12.5

15.7

18.0

42.8

TOTAL ALL

HOLLAND

IRELAND

ITALY

NORWAY

RUSSIA

SCOTLAND

SWEDEN

TURKEY

COUNTRIES

5.6

6.6

129.9

18.3

24.1

11.1

20.1

15.4

50.5

6.6

5.5

45.4

9.8

16.4

6.0

8.0

17.6

30.1

6.1

5.9

24.3

5.9

9.3

5.5

4.2

9.2

19.4

5.5

6.8

39.8

10.0

17.0

8.1

7.8

22.5

33.6

8.5

11.9

68.9

26.6

21.3

15.1

19.6

23.5

39.8

6.4

7.3

55.8

12.6

17.4

8.9

10.5

17.2

33.8

8.7

10.4

33.2

19.9

9.3

17.2

11.5

15.4

25.7

10.9

14.7

29.7

33.6

18.5

23.1

15.1

12.2

24.9

19.5

15.6

194.5

15.2

69.9

39.5

14.5

12.7

62.2

12.1

15.1

215.6

26.2

66.8

50.1

22.6

1.2

43.4

10. 1

19.1

36.3

35.1

46.7

36.5

15.2

4.6

22.4

11.5

13.7

53.2

25.1

17.5

25.0

14.7

13.3

30.3

8.4 9.6 54.6 16.9 17.4 13.7 12.1 15.7

100 AMERICAN INTELLIGENCE

length of residence, does it represent an error in the method
of measuring inteUigence, or, looked at from another angle,
does it show the gradually decreasing inteUigence of the
more recent immigrants examined in the army?

The hypothesis of growth of intelligence with increasing
length of residence may be identified with the hypothesis
of an error in the method of measuring intelligence, for we
must assume that we are measuring native or inborn intel-
ligence, and any increase in our test score due to any other
factor may be regarded as an error. It is therefore necessary
to examine two hypotheses, viz., (1) a defect in the measur-
ing scale, and (2) a change in the character of the immi-
grants examined, in order to decide which is correct, or, in
case both factors are operative, to estimate quantitatively
the magnitude of one of the factors, so that allowance may
be made for that factor and the weight of the other factor
thus determined.

The most probable source of error in our measure of in-
telligence is that arising from the different types of exam-
ination. Examination alpha involves the use of English,
and the ability to use English is a function of intelligence
and education in its broadest sense. Examination beta in-
volves no English, and the tests can not be considered as
educational measures in any sense. The individual exami-
nations were adapted to the linguistic ability of the person
examined. We would therefore expect to find an error in
two places only, first, in the selection of men for alpha and
beta, and second in the relationship between alpha and beta
as expressed on the combined scale.

If all members of our five years of residence groups had
been given alpha, beta and individual examinations in
equal proportions, then all would have been treated alike,
and the relationship shown would stand without any pos-
sibility of error. But this is not the case. The actual per-

AMERICAN INTELLIGENCE 101

centage of men in each residence group taking each type of
examination was as follows :

LENGTH OB' RESIDENCE

ALPHA

BETA

STANFORD-
BINET

PERFORMANCE
SCALE

to 5 years . . . .

19%

68%

11%

6 to 10 years . . . .

26%

65%

11 to 15 years. . . .

41%

54%

16 to 20 years . . . .

66%

32%

Over 20 years

73%

26%

y2%

Recognizing a variation in the type of examination given,
our problem becomes that of determining whether or not
any injustice has been done by converting results from
these different types of examination into the combined
scale. If the language and educational factors account for
the rise in the average score on the combined scale with in-
creasing years of residence, then we should expect that the
contribution made to the combined scale score through ex-
amination beta would remain constant, and the contribution
from examination alpha would increase very rapidly. On
the other hand, if the rise is independent of the language
and educational factor and due to the greater native intel-
ligence of the older groups, we should expect the contribu-
tion from both types of examination to remain the same.
In other words, if the combined scale is accurate, the dis-
parity between proportions taking different types of ex-
aminations would make no difference in the final results.
At the same time we would not expect to find the average
scores on the combined scale made by way of alpha and
beta to be the same, for beta was given not only to those
who were illiterate, but also to the dull and stupid who
failed to make a good score on alpha.

Computing, then, the average score on the combined scale
made by way of examination alpha, beta, and the Stanford-
Binet (which includes the results from the performance

102 AMERICAN INTELLIGENCE

scale) by each of the five years of residence groups, we find
the following:

TEABS OF SCORES DERIVED

JV!^^.T^^ SCORES DERIVED SCORES DERIVED IT" , tiLt!

BESIDENCE FROM THE

** FROM ALPHA FROM BETA i^xvwm xxiji,

GROUPS .tx^xi^ ^Kj ^ STANFORD-BINET

to 5 years 13.46 (S.D. 2.47) 11.11 (S.D. 2.73) 9.99 (S.D. 2.21)

6 to 10 years. . . 13.57 (S.D. 2.45) 11.29 (S.D. 2.65) 9.86 (S.D. 2.22)

11 to 15 years. . . 13.91 (S.D. 2.25) 11.62 (S.D. 2.50) 10.19 (S.D. 2.09)

16 to 20 years. . . 14.31 (S.D. 2.19) 12.22 (S.D. 2.68)

Over 20 years... 14.56 (S.D. 2.32) 11.93 (S.D. 2.70)

There is a steady progress in the scores in examination
alpha from "0 to 5 yrs." up to "over 20 yrs.," the total
gain being 1.12 points on the combined scale. There is an
equal amount of progress in the scores from examination
beta, a gain of 1.11 points on the combined scale being made
in a shorter period of time, i. e. from "0 to 5 yrs." to "16
to 20 yrs." If the increase in the average score on the com-
bined scale from 11.41 to 13.82 were due to the language
and educational factor, then the gain should come from
alpha and not from beta, for alpha involves language and
(indirectly) education, and beta does not. We actually find
that the gain from each type of examination is about the
same. This indicates, then, that the five years of residence
groups are groups with real differences in native intellig-
ence, and not groups laboring under more or less of a lin-
guistic and educational handicap.

There remains but one hypothesis that might establish
the fact that the increase in the score on the combined scale
with increasing length of residence was due to an error in
the measuring scale, and that is the hypothesis that the
combined scale was constructed in such a fashion that it
penalized individuals born in non-English speaking coun-
tries. It will be remembered that the combined scale was
constructed from Group X, a special experimental group
to which were given all three types of examination. Group

AMERICAN INTELLIGENCE 103

X was composed of 1047 individuals all of whom were born
in English speaking countries.

Fortunately we are able to test the reliability of the
combined scale under the most severe conditions. On page
654 of Memoir XV, Table 166 shows the scores on both
alpha and beta of all individuals in Groups I, II and III
of the principal sample who had been given both alpha and
beta. This group includes 4893 cases. It is obvious that we
may figure these 4893 cases as either alpha cases or beta
cases and convert them into the combined scale either by
Table 159 (the alpha conversion table) or Table 162 (the
beta conversion table). This the army writers have done,
and the results are given in Table 167 on page 655.

It is found that when we treat the 4893 cases as alpha
cases the average score on the combined scale is 10.775
(S.D.I. 64). When we treat the same4893 cases as beta cases,
the average score on the combined scale is 12.158 (S. D.
2.63). The actual difference between the two averages is
1.383 (=1=0.0298). In commenting on this result, the army
writers state:

"At first glance these results seem rather startling, for
one might suppose that going from alpha (for a given num-
ber of cases) to the combined scale ought to yield the same
results as going from beta to combined scale. The facts are
quite the contrary. However, this difference in no wise dis-
credits the method. It must be remembered that in a group
of this sort there is a large percentage of illiterates; thus
the group no doubt includes a considerable proportion of
the cases who made unsatisfactory scores in alpha and
were recalled to beta not because of stupidity but because
of language difficulty. When they reached beta, they were
able to make scores more consistent with their ability. It
is precisely this element of the group that causes the dif-
ference in the two means on the combined scale. The same

104 AMERICAN INTELLIGENCE

fact explains the wide differences in the standard devia-
tions. The standard deviation of the combined scale dis-
tribution when reached by way of beta is larger than by
way of alpha. Here the difference is no doubt due to the
fact that in alpha both the stupid and the non-English
speaking piled up in the lower class intervals, while in beta
the stupid remained in the lower ranges and the more in-
telligent went higher, thus increasing the standard devia-
tion." (p. 655.)

It is also possible to study the effect of using different
conversion tables in the case of Group X, the special ex-
perimental group which was composed of individuals who
were born in English speaking countries. The army writers
report on page 645 the analysis of Group X in this manner.
The following averages and standard deviations are re-
ported for the different methods of treating the data :

A\'ERAGE S. D.

(1) Treating all 1047 cases as measured by alpha only . . . 13.82716 3.03940

(2) Treating all cases scoring less than 25 points in alpha

as beta cases and the remainder as alpha

cases 13.88606 3.03776

(3) Treating all cases scoring less than 50 points in alpha

as beta cases and the remainder as alpha

cases 13.94350 3.20690

(4) Treating all cases scoring less than 75 points in alpha

as beta cases and the remainder as alpha

cases 13.96981 3.30997

The approximate agreement of the averages derived by
the four different methods is considered by the army
writers to be a proof of the validity of the transformation
tables which they publish for converting alpha, beta, and
Stanford-Binet distributions into combined scale distri-
butions.

We thus have two extreme instances of the results of

AMERICAN INTELLIGENCE 105

treating the same groups in different ways. If we take all
cases in the principal sample to whom were given both
alpha and beta, we find a difference of 1.383 (0.0298) in
the combined scale score of the group when figured first as
alpha cases and then as beta cases. On the other hand,
since the combined scale was empirically derived from the
1047 cases in Group X, it makes very little difference how
we treat the results of that group. We must now determine
whether the distributions of alpha scores of the foreign
born groups most closely resemble the distribution of alpha
scores in Group X, or the distribution of the alpha scores
of the 4893 cases who had both alpha and beta. This ques-
tion is very easily answered by turning to Table 7, which
shows the alpha score distribution of the following groups :

(1) The 4893 cases who had both alpha and beta.

(2) The 1047 cases in the special experimental group.

(3) The 679 cases of foreign born who had been in this

country from to 5 years.

(4) The 1098 cases of foreign born who had been in this

country from 6 to 10 years.

(5) The 1777 cases of foreign born who had been in this

country from to 10 years, a group obtained by com-
bining (3) and (4).
All five distributions have been made comparable by re-
ducing them to the proportion in each 1000 scoring at
each class interval of the alpha examination. Figure 34
shows graphically the relationships between the 4893 alpha
and beta cases, the 1047 cases in the special experimental
group and the 1777 foreign born who had been in this
country for 10 years or less. In Figure 34 the horizontal
line indicates alpha scores and the vertical line shows the
proportion in each 1000.

A hasty survey of Table 7 shows that our group of 4893
cases is a very specially selected group. 97.3% of the alpha

Table No. 7

Distribution of alpha scores of (1) all cases in Groups I, II,
and III given both alpha and beta, (2) special experi-
mental group, (3) foreign born in U. S. to 5 years, (4)
foreign born in U. S. 6 to 10 years, (5) foreign born in
U. S. to 10 years. All distributions reduced to common
denominator of number per 1000 at each class interval.
Actual distributions may be found on pages 621, 654 and
701 of Memoir XV.

4893 CASES

1047 CASES

679 CASES

1098 CASES

1777 CASES

ALPHA

TAKING BOTH

IN SPECIAL

IN u. a.

IN U. 8.

CT,ASS

ALPHA AND

EXPERIMENTAL

TO 5

6 TO 10

TO 10

INTEBVAL8

BETA

GROUP

TEARS

185-189

....

180-184

"2"

175-179

....
....

....

170-174

"4'

"2"

165-169

160-164

155-159

"k'

150-154

145-149

140-144

....

135-139

130-134

125-129

....

120-124

115-119

110-114

*'i*

105-109

....

100-104

95-99

90-94

85-89

80-84

75-79

70-74

65-69

60-64

55-59

50-54

45-49

40-44

85-39

30-34

25-29

20-24

104

16-19

149

10-14

286

6-9

207

0-4

104

106

AMERICAN INTELLIGENCE 107

scores of the 4893 cases are below a total alpha score of 50.
On the other hand, the distributions of alpha scores in our
groups of foreign born very closely resemble the distribu-
tion of alpha scores of Group X, the special experimental
group on which the combined scale was based. If we com-
pute the percentage of cases in each residence group graded
on alpha falling below a total alpha score of 50, we find the
following:

PER CENT. GRADED BY ALPHA
YEARS OF RESIDENCE GROUP AND FALLING BELOW A TOTAL

ALPHA SCORE OF 50

to 5 years 52.4%

6 to 10 years 52.2%

11 to 15 years 48.7%

16 to 20 years 39.1%

Over 20 years. 36.0%

Group X shows 47.7% of the 1047 cases falling below a
score of 50 on alpha. This group is approximately the same
as the group of foreign born we are studying. On the other
hand, the group of 4893 cases shows 97.3% of the alpha
scores below 50, and a very marked piling up of cases below
25 on alpha. The relationships between the distributions
of scores in the three groups are very clearly shown in
Figure 34.

It is clear that the combined scale would penalize the
foreign born only if all individuals who took alpha and
beta had been scored as alpha cases. As a matter of fact,
the opposite is true, for every individual who took both
alpha and beta was scored as a beta case, and the alpha
score was disregarded. On this account our results on the
foreign born are not subject to the distortion shown by
treating 4893 alpha and beta cases as alpha cases.

The 4893 cases treated as beta cases, not as alpha cases,
are in our group of 93,955 cases representing the white

FOREIGN BORN 0-10 YEARS IN U. S.

109

Figure 34. Distributions of alpha scores of three groups: (A)
foreign born individuals in this country from to 10 years, (B)
group X, the special experimental group, (C) all cases who took
both alpha and beta. The horizontal scale shows alpha scores,
and the vertical scale proportions in each thousand. The distri-
butions are drawn from the figures given in Table 7. To prove
that the combined scale does not penalize the foreign born, it is
only necessary to show that the distribution of alpha scores for
the foreign born is approximately similar to that of group X,
and unlike that of all alpha-beta cases. This similarity is appar-
ent from our chart. The men given both alpha and beta would
have been penalized if they had been scored through the com-
bined scale as alpha cases. This did not happen, however, for
the alpha scores of these men were disregarded, and only the
beta record used.

110 AMERICAN INTELLIGENCE

draft, and the chances are that most of them are among
the 12,492 members of this group who were born in foreign
countries. By treating the group as alpha cases, we show
that the maximum correction factor that it would be neces-
sary to apply to a group graded wrongly by alpha would be
1.383 points on the combined scale. But inasmuch as all
alpha and beta cases have been correctly graded by using
the beta conversion tables, there remain only those cases
graded by alpha alone who should have been given exam-
ination beta as well. It is of course impossible to estimate
the number of cases of this sort. If we make the very ex-
travagant assumption that 25% of the cases graded by
alpha alone in the residence group "0 to 5 yrs." should have
been graded by beta, and credit each of these cases with
1.383 points, the final correction that it would be necessary
to apply to our average would be 0.064 points, a quantity
which is negligible in view of the magnitude of the differ-
ences under consideration. Furthermore, we should not be
justified in using a correction factor of any sort unless it
could be shown that the distribution of the alpha scores
of the foreign born groups was very unlike that of the dis-
tribution of alpha scores of the special experimental group.
As a matter of fact, these distributions as shown in Figure
34 are very much alike.

We have therefore demonstrated the accuracy of the
combined scale as a measure of the intelligence of the
groups under consideration. We must therefore accept the
conclusion that under the conditions of this experiment the
differences shown in the average scores of the five years
of residence groups indicate real differences in intelligence
and not a defect in the measuring scale. Instead of con-
sidering that our curve (Figure 33) indicates a growth of
intelligence with increasing length of residence, we are
forced to take the reverse of the picture and accept the

AMERICAN INTELLIGENCE 111

hypothesis that the curve indicates a gradual deterioration
in the class of immigrants examined in the army, who came
to this country in each succeeding ^ve year period since
1902.

SECTION V

ANALYSIS OF IMMIGRATION TO THE
UNITED STATES

The fact that the average inteUigence of the immigrants
examined in the army who came to this country in each
successive five year period since 1902 becomes progres-
sively lower with each succeeding period indicates that an
explanation of this phenomenon might be found in a change
in the character of immigration. We must, therefore, turn
to the statistics on immigration to see if any such change
can be detected. Table 8 shows the percentage of the total
immigration coming from various countries in the periods
roughly corresponding to the five years of residence periods
covered in the army statistics. The data on immigration
were obtained from the Statistical Abstract of the United
States for the years 1900, 1910, and 1920. Table 8 reads
as follows: in the years 1887 to 1897, the period roughly
corresponding with our residence group "over 20 yrs.,"
10.9% of our total immigration came from England, 2.7%
from Scotland, 1% from Holland, etc. The relations shown
in Table 8 are shown graphically in Figure 35. Each com-
plete bar in Figure 35 represents 100%. The per cent, which
each country has contributed to the total immigration of
each period has been scaled off proportionately in each bar.
These figures show that the most abrupt change in the
character of immigration came between the periods 1887-
1897 and 1898-1902. These periods show a very marked
decrease in the proportion of the immigration from Eng-
land and Germany, and a substantial decrease in the pro-
portion of immigration from Scotland, Sweden, and

112

AMERICAN INTELLIGENCE 113

Table No. 8

Per cent, of total immigration coming from various coun-
tries during periods roughly corresponding to the five
years of residence groups.

1913-1917 1908-1912 1903-1907 1898-1902 1887-1897
TO 5 6 TO 10 11 TO 15 16 TO 20 over 20

YEARS YEARS YEARS YEARS YEARS

England 3.7 5.1 4.6 2.6 10.9

Scotland 1.0 1.8 1.4 0.5 2.7

Holland 0.6 0.8 0.5 0.4 1.0

Germany 2.5 3.5 4.0 4.8 18.7

Denmark 0.7 0.7 0.8 0.8 1.7

Canada 13.5 6.0 0.6 0.2 No record

Sweden 1.5 2.0 2.8 4.6 7.3

Norway 1.1 1.5 2.3 2.4 2.7

Belgium 0.5 0.5 0.5 0.3 0.6

Ireland 2.4 3.3 3.8 7.2 11.9

Austria 16.7 21.8 24.9 23.6 11.6

Turkey 2.6 3.4 1.9 0.7 No record

Greece 3.7 2.5 1.8 1.1 0.2

Russia 17.8 18.3 18.3 17.8 12.0

Italy 20.0 20.2 23.7 25.8 12.4

Allothers 11.7 8.6 8.1 7.2 6.3

Ireland. On the other hand, the proportion of immigrants
coming from Austria, Russia, and Italy showed a marked
increase at this time. In general the following relations
held:

England showed a decided drop in the proportion of im-
migrants furnished between the period 1887-1897 and the
period 1898-1902. There has been a slight increase since
1898, but the proportion is less than 5%, when formerly it
was over 10%.

Scotland contributed 2.7% of our total immigration in
the period 1887-1897, and since that time, never more than

Holland never contributed more than 1% of our total
immigration in any period covered by these figures.

Germany contributed 18.7% of our immigration in the
period 1887-1897, 4.8% in the period 1898-1902, and since
that time the proportion has decreased with each succeed-
ing period.

DATES 1887

1897

YEARS ovEn

ENGLAND

SCOTLAND
HOLLAND

GERMANY

DENMARK
SWEDEN

NORWAY
BELGIUM

IRELAND

AUSTRIA

GREECE
RUSSIA

ITALY

ALL OTHERS

TO
1912

6 10

^ -=;^^- - ^ ^ GERMAN Y

-^-^^.^_ _ r^^^-.^-c::-p DENMARK

1918 DATES

1917

0-5 YEARS
~JH| ENGLAND
. - SCOTLAND
rteBHOLLAND
^m GER>IAN

CANADA

SWEDEN
NORWAY
BELGIUM
IRELAND

AUSTRIA

M TURKEY
GREECE

RUSSIA

L^"i

ITALY

ALL OTHERS

114

Figure 35. Analysis of immigration by countries. Each vertical
bar represents 100%, and each subdivision represents the per
cent, of the total immigration that each country furnished in the
period under consideration. It is apparent that the most sudden
change in the character of immigration came between the per-
iods 1887-1897 and 1898-1902. On the other hand, the curve
showing the relationship between years of residence and average
intelligence shows no correspondingly large difference in the in-
telligence of immigrants who came here in these periods (Figure
22, Groups "16 to 20" and "over 20"). Figure 22 shows a large
difference between the "16 to 20 yrs." and "11 to 15 yrs."
groups, but the chart above shows no marked difference in the
immigration coming to this country in the corresponding five-
year periods, 1898-1902 and 1903-1907.

115

116 AMERICAN INTELLIGENCE

Denmark contributed 1.7% of our immigration in the
period 1887-1897 and since that time less than 1% in each
period.

The immigration from Canada was not reported prior to
1896. There was a very marked increase in the percentage
of our total immigration coming from Canada in the period
1908-1912 and again in the period 1913-1917.

The proportion of immigration coming from Sweden has
decreased steadily from 7.3% in the period 1887-1897 to
1.5% in the period 1913-1917. A similar decrease from
2.7% to 1.1% is shown by Norway during the same periods.

Belgium has contributed less than 1% to the total immi-
gration in each period under consideration.

The immigration from Ireland has decreased in propor-
tion from 11.9% to 2.4% in the five year periods shown.

All other countries show a gain in the proportion of im-
migrants which they supply. Austria supplied 11.6% in the
period 1887-1897, this proportion jumping to 23.6% in the
period 1898-1902 and remaining above 20% until the last
period 1913-1917, a period which reflects the war conditions
in Europe.

The proportion of immigration from Russia has increased
from 12% prior to 1898 to 18% since that time. The pro-
portion of immigration from Italy, which was about 12%
prior to 1898, has never been below 20% since. Turkey and
Greece show a small but increasing proportion in the suc-
cessive periods covered.

Enough evidence has been cited to show that there has
been some change in the character of our immigration dur-
ing the periods covered in the army report. The gradual
decline in the average inteUigence of the more recent im-
migrants examined in the army might be due to these
changes in the source of supply. If this hypothesis is correct,
we should expect to find differences between the scores of

AMERICAN INTELLIGENCE 117

the representatives of each country making up our total
group of the foreign born white draft.

SECTION VI

ANALYSIS OF THE FOREIGN BORN WHITE
DRAFT BY COUNTRY OF BIRTH

The army reports show distributions of psychological
test scores for all of the 12,492 foreign born cases classified
according to the country of birth, although these figures
for each country are not sub-divided again into years of
residence groups. As a matter of fact, when we break up
the 12,492 cases according to the country of birth, the
figures from certain countries become very small; so that
further sub-division would make the results valueless. We
can, however, examine the figures which give us an intelli-
gence measure of the foreign born men of our army and take
them for what they are worth. Even though the number
of cases is very small for certain nationalities, we previous-
ly saw that the reliability of the difference does not depend
entirely on the number of cases, but on three factors the
size of the difference, the variability of each series of meas-
urements, and the number of cases in each series.

The same tables from which we derived the 12,492 for-
eign born cases (Tables 207, 208, 209 and 210; pp. 692, 693
and 694) give the classification of test scores by country of
birth. From these tabulations the combined scale score of
each nativity group has been computed. The actual dis-
tributions of these combined scale scores are shown in
Table 9. These distributions reduced to percentages are
shown in Table 10.

The differences found, expressed in terms of the per cent,
from each country who exceed the average native born
American, are as follows:

118

AMERICAN INTELLIGENCE

119

England 67.3%

Belgium

35.3%

Scotland 58.8%

Austria

28.2%

Holland 58.1%)

Ireland

26.2%

Germany 48.7%

Turkey

25.3%

Denmark 47.8%

Greece

21.3%

Canada 47.3%

Russia

18.9%

Sweden 41.7%

Italy

14.4%

Norway 37.3%

Poland

12.2%

|. The actual diifferences between the average scores on the
combined scale of each country and every other country,
together with the probable errors of the differences, and
the ratios of the probable errors of the differences to the
differences are shown in Tables 11 to 27 inclusive. In these
tables, the convention has been followed of marking a dif-
ference "unreliable" if the actual difference was less than
four times the probable error of the difference. The coun-
tries tabulated are arranged in the order of the increasing
reliability of the differences found, above and below the
average of the country with which they are compared.

The relations between the averages shown in Table 9 are
shown graphically in Figure 36. For the convenience of
those who use Stanford-Binet "mental ages," one side of the
scale in Figure 36 has been drawn so as to read in "mental
ages," and the other side to read in combined scale units
the units in which our averages and measures of variability
have been calculated. The Stanford-Binet "mental age"
scale was calculated from the regression equation given on
page 654 of Memoir XV:

Mental age (in years) = 0.778 C.S.-|-2.606.

Table No. 9

Analysis of foreign born white draft by country of birth.
Distribution of scores on the combined scale, of men
born in each country.

COMBINED

SCALE

ENGLAND

SCOTLAND

HOLLAND

GERMANY

DENMARK

CANADA

SWEDEN

NORWAY

INTERVALS

24.0-24.9

0.1

23.0-23.9

0.7

....

22.0-22.9

1.7

6.5

o.i

0.5

0.6

21.0-21.9

1.4

0.7

1.2

'o'.i

4.3

0.5

o'.i

20.0-20.9

6.0

1.9

1.6

2.4

0.5

8.7

2.2

1.2

19.0-19.9

10.7

2.6

3.0

5.1

1.9

15.8

5.4

3.2

18.0-18.9

21.1

3.7

4.5

7.1

5.8

28.2

10.2

8.4

17.0-17.9

33.7

8.9

7.2

14.7

13.7

47.8

21.3

18.4

16.0-16.9

48.9

17.9

12.9

24.7

26.4

77.2

44.3

34.0

15.0-1.5.9

64.2

20.8

20.4

34.4

39.9

104.4

72.6

59.1

14.0-14.9

71.5

22.9

25.1

46.3

52.0

135.2

101.9

81.4

13.0-13.9

65 7

23.0

25.5

57.1

64.0

163.2

125.6

98.5

12.0-12.9

40.9

18.0

18.6

53.2

54.2

144.4

118.3

101 2

11.0-11.9

19.2

11.0

10.2

31.5

31.2

93.5

79.3

79.5

10.0-10.9

10.2

6.1

5.8

16.4

18.2

60.2

51.4

55.8

9.0-9.9

6.1

3.5

2.8

8.0

10.3

41.7

31.7

36.4

8.0-8.9

3.7

2.1

1.1

3.5

4.6

25.0

16.2

19.8

7.0-7.9

2.2

1.1

0.4

1.3

1.7

13.0

7.0

9.2

6.0-6.9

1.0

0.5

0.1

0.5

0.4

5.6

2.4

3.4

5.0-5.9

0.3

0.1

2.1

0.6

1.0

4.0-4.9

0.1

0.8

0.1

S.0-3.9

'.'.'.'.

0.2

2.0-2.9

0.1

....

1.0-1.9

....

No. cases

411

146

140

308

325

972

691

611

Average

14.87

14.34

14.32

13.88

13.69

13.66

13.30

12.98

S. D..

2.57

2.63

2.39

2.43

2.23

2.67

2.38

2.47

150

BELGIUM

IHF.T.AND

AUSTRIA

TURKEY

GREECE

RUSSIA

ITALY

POLAND

6!i

O.Q

b'.k

1.1

'6'.4

0.3

....

2.3

0.8

2.0

1.4

0.8

0.2

3.8

2.1

o.k

3.3

2.0

1.5

0.8

7.7

5.9

o'.k

0.8

7.5

3.0

4.3

3.0

20.7

14.4

0.9

2.4

13.5

5.0

9.6

8.2

40.0

32.8

2.7

7.0

23.7

10.1

15.6

14.8

62.7

70.0

5.8

13.3

38.2

20.1

24.4

26.5

97.7

136.4

11.1

18.6

59.4

32.1

36.9

49.8

152.8

235.0

18.3

22.3

102.0

46.8

57.2

83.0

233.6

377.2

33.1

22.6

122.6

50.8

67.4

99.2

301.1

508.7

47.2

16.5

94.0

38.2

56.9

88.5

317.0

569.2

53.8

11.0

68.9

29.3

47.8

73.1

316.7

596.3

56.0

7.3

55.0

24.7

41.0

57.3

299.2

573.6

55.2

4.4

35.7

17.5

28.9

35.5

226.0

423.4

43.4

2.6

18.6

10.6

17.3

19.0

141.3

255.7

28.5

1.3

8.0

5.4

8.2

8.5

71.1

126.6

15.2

0.5

3.0

2.3

3.3

3.2

29.9

53.0

6.8

0.2

1.0

0.9

1.2

1.0

11.2

19.5

2.6

0.3

3.4

6.0

0.9

.'.'.

0.1

1.0
0.2

1.8
0.4

0.3

129

658

301

423

572

2340

4009

382

12.79

12.32

12.27

12.02

11.90

11.34

11.01

10.74

2.42

2.60

2.75

2.45

2.83

2.60

2.59

121

Table No. 10

Analysis of foreign born white draft by country of birth.
Per cent, from each country scoring at each interval on
the combined scale.

COMBINED

SCALE

ENGLAND

SCOTLAND

HOLLAND

DENMABE

CANADA

SWEDEN

NOBWAY

miEBVALS

23.0-23.9

0.2

....

22.0-22.9

0.4

0.3

'6!i

0.2

o'.i

21.0-21.9

0.7

1.0

0.5

0.4

....

0.4

'6!i

....

20.0-20.9

1.5

1.3

1.1

0.8

'6!2

0.9

0.3

*6.2

19.0-19.9

2.6

1.8

2.1

1.7

0.6

1.6

0,8

0.5

18.0-18.9

5.1

2.5

3.2

2.3

1.8

2.9

1.6

1.4

17.0-17.9

8.2

6.1

4.8

4.2

4.9

3.1

3.0

16.0-16.9

11.9

12.2

9.3

8.0

8.2

7.9

6.4

5.6

16.0-15.9

15.6

14.3

14.6

11.2

12.3

10.8

10.6

9.7

14.0-14.9

17.4

15.7

17.9

15.0

16.0

13.9

14.8

13.3

13.0-13.9

16.0

15.8

18.2

18.5

19.7

16.8

18.1

16.1

12.0-12.9

10.0

12.3

13.3

17.3

16.6

14.9

17.2

16.6

11.0-11.9

4.7

7.5

7.3

10.2

9.6

11.4

13.0

10.0-10.9

2.5

4.2

4.1

5.3

5.6

6.2

7.6

9.1

9.0-9.9

1.5

2.4

2.0

2.6

3.2

4.3

4.6

6.0

8.0-8.9

0.9

1.4

0.8

1.1

1.4

2.6

2.8

3.8

7.0-7.9

0.5

0.8

0.3

0.4

0.5

1.3

1.0

1.6

6.0-6.9

0.2

0.4

0.1

0.2

0.1

0.8

0.3

0.6

5.0-5.9

0.1

0.2

0.1

0.2

4.0-4.9

0.1

....

S.0-3.9

....

2.0-2.9

Ifd

BELGIUM

IBELAND

AU8TBIA

TUEKET

GREECE

RUSSIA

ITALY

POLAND

'. . . .

6.2

'6!i

o'.i

'6!i

....

0,3

0.5

0.2

....

0.2

'0.2

0.5

0.7

0.4

'6!i

0.4

'6!i

'6]i

0.6

1.1

1.0

0.5

0.9

0.4

0.2

1.9

2.1

1.7

2.3

1.4

1.7

0.8

0.7

5.4

3.6

3.4

3.7

2.6

1.7

1.5

10.3

5.8

6.6

5.8

4.7

4.2

3.4

2.9

12.9

9.0

10.6

8.7

6.5

5.8

4.8

17.3

15.5

15.6

13.5

14.5

10.1

9.4

8.7

17.5

18.7

16.9

15.9

17.4

12.9

12.7

12.3

12.8

14.3

12.7

13.4

15.4

13.5

14.2

14.1

8.5

10.5

9.7

11.3

12.8

13.5

14.9

14.6

5.6

8.3

8.2

9.7

10.0

12.8

14.3

14.4

3.4

5.4

5.8

6.8

6.2

9.7

10.5

11.4

2.0

2.8

3.5

4.1

3.3

6.0

6.4

7.5

1.0

1.2

1.8

1.9

1.5

3.0

3.1

4.0

0.4

0.5

0.8

0.6

1.3

1.7

1.8

0.2

0.3

0.2

0.5

0.7

0.1

0.2
0.1

123

USCOfSccnl IQ^

Enghnd 14^

Scdhnd 1434-^
Holland 14^^

U.STwhitcJ 13.77 -

DanmarK I5e9 <^
Canada 1566 ^
^*^den 030 -

Norv^ay I2.S8 -
Belgium 18.79 -

Jrvhnd IZZe ^
Austria I^.ZT^

TUrKeyl^OZ-
Greece 1130 -

le-

/^sjia II.S4
Italy 11.01
AAtnd 10.74
US(Cohnd) I04t

11-

I 10-

1^4

Figure 36. The relative standing of the nativity groups according
to their average intelligence. The averages of the nativity groups
are taken from Table 9. The averages of the vrhite officers and
negro draft (from Table 1) and the native born white draft
(from Table 2) are also shown. The left hand scale reads in
units of the combined scale. The right hand scale reads in
units of "mental age" representing what would be the approxi-
mately equivalent scores on the Stanford revision of the Binet-
Simon scale. In interpreting the differences shown, it must be
remembered that all the differences are not equally reliable, for
the reliability of the measurements depends on the number of
cases in each group and the variability of the group. The relia-
bility of all the differences is shown in Tables 11 to 27.

12A

Table No. 11

Differences between ENGLAND and other countries

Number of cases 411
Average score 14.87
Standard deviation 2.57

Scotland

-0.53 0.1695(3.1) Difference unreliable

Holland

-0.55 0.1606(3.4) Difference unreliable

Germany

-0.99 0.1264 (7.8)

Denmark

-1.18 0.1192(9.9)

Canada

-1.21 0.1030(11.7)

Belgium

-2.08 0.1669(12.5)

United States

-1.10 0.0855(12.9)

Sweden

-1.57 0.1049(14.9)

Norway

-1.89 0.1087(17.4)

Austria

-2.60 0.1368 (19.0)

Turkey

-2.85 0.1241 (22.9)

Ireland

-2.55 0.1094 (23.3)

Greece

-2.97 0.1097(27.0)

Poland

-4.13 0.1236(33.4)

Russia

-3.53 0.0940(37.5)

Italy

-3.86 0.0897(43.0)

126

Table No. 12

Differences between SCOTLAND and other countries

Number of cases 146
Average score 14.34
Standard deviation 2.63

England

Holland

Germany

United States

Denmark

Canada

Sweden

Belgium

Norway

Austria

Ireland

Turkey

Greece

Russia

Poland

Italy

+0.53
-0.02
-0.46
-0.57
-0.65
-0.68
-1.04
-1.55
-1.36
-2.07
-2.02
-2.32
-2.44
-3.00
-3.60
-3.33

0.
0,

.1695 (3.1)
.1999 (0.1)
.1736 (2.6)
.1466 (3.9)
.1685 (3.9)
.1575 (4.3)
.1587 (6.5)
.2050 (7.6)
.1612 (8.4)
.1813 (11.4)
.1617 (12.5)
.1720 (13.5)
.1619 (15.0)
.1517 (19.8)
.1716 (20.9)
.1491 (22.3)

Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.

127

Table No. 13

Differences between HOLLAND and other countries

Number of cases 140
Average score 14.32
Standard deviation 2.39

England

Scotland

Germany

Denmark

United States

Canada

Sweden

Belgium

Norway

Austria

Ireland

Turkey

Greece

Russia

Poland

Italy

+0.55
+0.02
-0.44
-0.63
-0.55
-0.66
-1.02
-1.53
-1.34
-2.05
-2.00
-2.30
-2.42
-2.98
-3.58
-3.31

=tO
0,

1606 (3.4)
1999 (0.1)
1649 (2.7)
1595 (3.9)
1362 (4.0)
1479 (4.4)
1491 (6.8)
1977 (7.7)
1518 (8.8)
1731 (11.8)
1523 (13.1)
1632 (14.1)
1525 (15.9)
1417 (21.0)
1628 (21.9)
1388 (23.9)

Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.

1^8

Table No. 14

Differences between GERMANY and other countries

Number of cases 308
Average score 13.88
Standard deviation 2.43

England
Holland
Scotland
United States
Denmark
Canada
Sweden
Belgium
Norway-
Austria
Ireland
Turkey
Greece
Poland
Russia
Italy

+0.99
+0.44
+0.46
-0.11
-0.19
-0.22
-0.58
-1.09
-0.90
-1.61
-1.56
-1.86
-1.98
-3.14
-2.54
-2.87

0.1264
0.1649
0.1736
0.0934
0.1249
0.1097
0.1114
0.1711
0.1150
0.1418
0.1157
0.1297
0.1159
0.1291
0.1012
0.0972

(7.8)

(2.7)

(2.6)

(1.1)

(1.5)

(2.0)

(5.2)

(6.3)

(7.8)

(11.3)

(13.5)

(14.3)

(17.1)

(24.3)

(25.1)

(29.5)

Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.

129

Table No. 15

Differences between the UNITED STATES and other
countries. (Native born white draft used.)

Number of cases 81,465
Average score 13.77
Standard deviation 2.86

England

Holland

Scotland

Germany

Denmark

Canada

Belgium

Sweden

Norway

Austria

Turkey

Ireland

Greece

Poland

Russia

Italy

+ 1.10
+0.55
+0.57
+0.11
-0.08
-0.11
-0.98
-0.47
-0.79
-1.50
-1.75
-1.45
-1.87
-3.03
-2.43
-2.76

0.0855
0.1362
0.1466
0.0934
0.0835
0.0582
0.1437
0.0614
0.0678
0.1071
0.0904
0.0688
0.0693
0.0896
0.0400
0.0285

(12.9)

(4.0)

(3.9)

(1.1)

(1.0)

(1.9)

(6.8)

(7.6)

(11.6)

(14.0)

(19.3)

(21.2)

(26.9)

(33.8)

(60.7)

(96.8)

Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.

ISO

Table No. 16
Differences between DENMARK and other countries

Number of cases 325

Average score 13.69

Standard deviation 2.23

England

+ 1.18 0.1192(9.9)

Scotland

+0.65 0.1685(3.9) Difference unreliable.

Holland

+0.63 0.1595(3.9) Difference unreliable,

Germany

+0.19 0.1249(1.5) Difference unreliable.

United States +0.08 0.0835 (1.0) Difference unreliable,

Canada

-0.03 0.1013(0.3) Difference unreliable.

Sweden

-0.39 0.1032(3.7) Difference unreliable.

Belgium

-0.90 0.1659(5.4)

Norway

-0.71 0.1071 (6.6)

Austria

-1.42 0.1355(10.5)

Ireland

-1.37 0.1079(12.7)

Turkey

-1.67 0.1228(13.6)

Greece

-1.79 0.1081(16.5)

Poland

-2.95 0.1221 (24.1)

Russia

-2.35 0.0921 (25.5)

Italy

-2.68 0.0877(30.5)

131

Table No. 17
Differences between CANADA and other countries

Number of cases 972
Average score 13.66
Standard deviation 2.67

England

+ 1.21 0.1030(11.7)

Holland

+0.66 0.1479(4.4)

Scotland

+0.68 0.1575(4.3)

Germany

+0.22 0.1097(2.0) Difference unreliable,

United States +0.11 0.0582(1.9) Difference unreliable,

Denmark

+0.03 0.1013(0.3) Difference unreliable.

Sweden

-0.36 0.0840(4.3)

Belgium

-0.87 0.1547(5.6)

Norway

-0.68 0.0888(7.6)

Austria

-1.39 0.1215(11.4)

Ireland

-1.34 0.0896(14.9)

Turkey

-1.64 0,1071(15.3)

Greece

-1.76 0.0900(19.6)

Poland

-2.92 0.1064(27.4)

Russia

-2.32 0.0700(33.1)

Italy

-2.65 0.0641 (41.3)

132

Table No. 18
Differences between SWEDEN and other countries

Number of cases 691
Average score 13.30
Standard deviation 2.38

England

+ 1.57 0.1049(14.9)

United States +0.47 0.0614 (7.6)

Holland

+ 1.02 0.1491(6.8)

Scotland

+ 1.04 0.1587(6.5)

Germany

+0.58 0.1114(5.2)

Canada

+0.36 0.0840(4.3)

Denmark

+0.39 0.1032(3.7) Difference unreliable,

Belgium

0.51 0.1559(3.3) Difference unreliable,

Norway

-0.32 0.0910(3.5) Difference unreliable.

Austria

-1.03 0.1231(8.4)

Ireland

-0.98 0.0918(10.6)

Turkey

-1.28 0.1089(11.8)

Greece

-1.40 0.0921(15.2)

Poland

-2.56 0.1082(23.6)

Russia

-1.96 0.0727(26.9)

Italy

-2.29 0.0671(34.1)

133

Table No. 19
Differences between NORWAY and other countries

Number of cases 611
Average score 12.98
Standard deviation 2.47

England

+ 1.89 =i= 0.1087 (17.4)

United States +0.79 0.0678 (11.6)

Holland

+ 1.34 0.1518(8.8)

Scotland

+ 1.36 0.1612(8.4)

Germany

+ 0.90 0.1150(7.8)

Canada

+0.68 0.0888(7.6)

Denmark

+0.71 0.1071(6.6)

Sweden

+0.32 0.0910(3.5) Difference unreliable

Belgium

-0.19 0.1586(1.2) Difference unreliable

Austria

-0.71 0.1264(5.6)

Ireland

-0.66 0.0961 (6.8)

Turkey

-0.96 0.1126(8.5)

Greece

-1.08 0.0965(11.2)

Poland

-2.24 0.1119(20.1)

Russia

-1.64 0.0795(20.6)

Italy

-1.97 0.0729(27.0)

134

Table No. 20

Differences between BELGIUM and other countries

Number of cases 129
Average score 12.79
Standard deviation 2.42

England

Holland

Scotland

United States

Germany

Canada

Denmark

Sweden

Norway

Ireland

Austria

Turkey

Greece

Russia

Poland

Italy

+2.08
+ 1.53
+ 1.55
+0.98
+ 1.09
+0.87
+0.90
+0.51
+0.19
-0.47
-0.52
-0.77
-0.89
-1.45
-2.05
-1.78

==0

=*=o

,1669 (12.5)
,1977(7.7)
,2050 (7.6)
1437 (6.8)
1711 (6.3)
,1547 (5.6)
1659 (5.4)
,1559 (3.3)
,1586 (1.2)
1590 (2.9)
,1789 (2.9)
1695 (4.5)
1592 (5.6)
,1488 (9.8)
,1691 (12.1)
,1462 (12.2)

Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.

135

Table No. 21
Differences between IRELAND and other countries

Number of cases 658
Average score 12.32
Standard deviation 2.60

England

+2.55 0.1094(23.3)

United States +1.45 0.0688 (21.2)

Canada

+ 1.34 0.0896(14.9)

Germany

+ 1.56 0.1157(13.5)

Holland

+2.00 0.1523(13.1)

Denmark

+ 1.37 0.1079(12.7)

Scotland

+2.02 0.1617(12.5)

Sweden

+0.98 0.0918(10.6)

Norway

+0.66 0.0961 (6.8)

Belgium

+0.47 0.1590(2.9) Difference unreliable

Austria

0.05 0.1269(0.4) Difference unreliable

Turkey

0.30 0.1132(2.7) Difference unreliable

Greece

-0.42 0.0972(4.3)

Russia

-0.98 0.0791 (12.4)

Poland

-1.58 0.1126(14.0)

Italy

-1.31 0.0739(17.7)

1S6

Table No. 22

Differences between AUSTRIA and other countries

Number of cases 301
Average score 12.27
Standard deviation 2.75

England

United States

Holland

Canada

Scotland

Germany

Denmark

Sweden

Norway

Belgium

Ireland

Turkey

Greece

Russia

Poland

Italy

+2.60
+ 1.50
+2.05
+ 1.39
+2.07
+ 1.61
+ 1.42
+ 1.03
+0.71
+0.52
+0.05
-0.25
-0.37
-0.93
-1.53
-1.26

0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.

1368 (19.0)
1071 (14.0)
1731 (11.8)
1215 (11.4)
1813 (11.4)
1418 (11.3)
1355 (10.5)
1231 (8.4)
1264 (5.6)
1789 (2.9)
1269 (0.4)
1398 (1.8)
1272 (2.9)
1139 (8.2)
1393 (10.9)
1104 (11.4)

Difference unreliable.
Difference unreliable.
Difference unreliable.
Difference unreliable.

137

Table No. 23

Differences between TURKEY and ot^er countries

Number of cases 423
Average score 12.02
Standard deviation 2.75

England

+2.85 0.1241 (22.9)

United States +1.75 0.0904 (19.3)

Canada

+ 1.64 0.1071 (15.3)

Germany

+ 1.86 0.1297(14.3)

Holland

+2.30 0.1632(14.1)

Denmark

+ 1.67 0.1228(13.6)

Scotland

+2.32 0.1720(13.5)

Sweden

+ 1.28 0.1089(11.8)

Norway

+0.96 0.1126(8.5)

Belgium

+0.77 0.1695(4.5)

Ireland

+0.30 0.1132(2.7) Difference unreliable

Austria

+0.25 0.1398(1.8) Difference unreliable

Greece

-0.12 0.1135(1.0) Difference unreliable

Russia

-0.68 0.0984(6.9)

Poland

-1.28 0.1269(10.1)

Italy

-1.01 0.0944(10.7)

138

Table No. 24
Differences between GREECE and other countries

Number of cases 572
Average score 11.90
Standard deviation 2.45

England

+2.97 0.1097(27.0)

United States +1.87 0.0693 (26.9)

Canada

+ 1.76 0.0900(19.6)

Germany

+ 1.98 0.1159(17.1)

Denmark

+ 1.79 0.1081(16.5)

Holland

+2.42 0.1525(15.9)

Sweden

+ 1.40 0.0921(15.2)

Scotland

+2.44 0.1619(15.0)

Norway

+ 1.08 0.0965(11.2)

Belgium

+0.89 0.1592(5.6)

Ireland

+0.42 0.0972(4.3)

Austria

+0.37 0.1272(2.9) Difference unreliable.

Turkey

+0.12 0.1135(1.0) Difference unreliable.

Russia

-0.56 0.0795(7.1)

Poland

-1.16 0.1129(10.3)

Italy

-0.89 0.0744(11.9)

139

Table No. 25
Differences between RUSSIA and other countries

Number of cases 2340
Average score 11.34
Standard deviation 2.83

United States +2.43 0.0400 (60.7)
England +3.53 0.0940 (37.5)

Canada +2.32 0.0700 (33.1)

Sweden +1.96 0.0727 (26.9)

Denmark +2.35 0.0921 (25.5)
Germany . +2.54 0.1012 (25.1)

Holland +2.98 0.1417 (21.0)

Norway +1.64 0.0795 (20.6)

Scotland +3.00 0.1517 (19.8)

Ireland +0.98 0.0791 (12.4)

Belgium +1.45 0.1488 (9.8)

Austria +0.93 0.1139 (8.2)

Greece +0.56 0.0795 (7.1)

Turkey +0.68 0.0984 (6.9)

Poland -0.60 0.0977 (6.1)

Italy -0.33 0.0483 (6.9)

140

Table No. 26
Differences between ITALY and other countries

Number of cases 4009
Average score 11.01
Standard deviation 2.60

United States +2.76 0.0285 (96.8)

England +3.86 0.0897 (43.0)

Canada +2.65 0.0641 (41.3)

Sweden +2.29 0.0671 (34.1)

Denmark +2.68 0.0877 (30.5)

Germany +2.87 0.0972 (29.5)

Norway +1.97 0.0729 (27.0)

Holland +3.31 0.1388(23.9)

Scotland +3.33 0.1491 (22.3)

Ireland +1.31 0.0739(17.7)

Belgium +1.78 0.1462 (12.2)

Austria +1.26 0.1104 (11.4)

Turkey +1.01 0.0944(10.7)

Greece +0.89 0.0744 (11.9)

Russia +0.33 0.0483 (6.9)

Poland -0.27 0.0936(2.9) Difference unreliable.

141

Table No. 27
Differences between POLAND and other countries

Number of cases 382
Average score 10.74
Standard deviation 2.59

United States +3.03 0.0896 (33.8)

England +4.13 0.1236 (33.4)

Canada +2.92 0.1064 (27.4)

Germany +3.14 0.1291 (24.3)

Denmark +2.95 0.1221 (24.1)

Sweden +2.56 0.1082 (23.6)

Holland +3.58 0.1628 (21.9)

Scotland +3.60 0.1716 (20.9)

Norway +2.24 0.1119 (20.1)

Ireland +1.58 0.1126 (14.0)

Belgium +2.05 0.1691 (12.1)

Austria +1.53 0.1393 (10.9)

Greece +1.16 0.1129 (10.3)

Turkey +1.28 0.1269 (10.1)

Russia +0.60 0.0977 (6.1)

Italy +0.27 0.0936(2.9) Difference unreliable.

AMERICAN INTELLIGENCE 143

Tables 11 to 27 give the most accurate interpretation of
the differences found between the various nativity groups
that it is possible to derive from the army data. It is desir-
able, however, to attempt to estimate the meaning of these
differences in terms of standards which have some popular
significance. For this reason, the combined scale distribu-
tions in this study have been converted into estimates of
the per cent, of A and B men in each group, and the per
cent, of D, D and E men.

It should be remembered that the army letter ratings
are arbitrary ratings and have no real significance aside
from the tests from which they were derived. The army
rating "A" represents a certain score on the tests that
should have been reached by 4% or 5% of the whole army
group. In the same way the rating "B" was fixed so as to
include the next 8% or 10%. It is of course absurd to deplore
the fact that only 4% or 5% of the army were A men, when
A was fixed so that only 4% or 5% could receive that rating.
At the other end of the scale, the ratings D and D were
fixed so that they would include approximately 20% of the
total group, and the E rating was reserved for those recom-
mended for development battalions, special service organi-
zations, rejection or discharge. The estimates made at the
time the examinations were being standardized proved to
be about right. The A and B groups which should have in-
cluded 12% to 15% of the draft actually included 12%, and
the D, D and E groups, which should have included 20%
to 25%, actually included 24%.

Table 28 gives the per cent, of cases in each nativity
group who would be classified A or B according to the
criterion of the upper 12% of the total unselected white
draft. Table 29 gives the per cent, who would be classified
as D, D and E according to the criterion of 24% of the
unselected group. The relations between the various nativ-

144 AMERICAN INTELLIGENCE

ity groups given in Tables 28 and 29 are shown graphically
in Figure 37.

Another criterion that probably represents intellectual
ability of a high order is the per cent, at or above the
average of the white officers. The classification of the
nativity groups according to this criterion is given in Table
30. At the other end of the scale, a criterion of inferiority
that has n certain social significance is the per cent, at or
below the average of the negro draft. Table 31 shows the
nativity groups classified according to this criterion. The
results given in Tables 30 and 31 are shown graphically in
Figures 38 and 39.

It is not possible to determine accurately the percentage

Table No. 28
Per cent, of each nativity group in the A and B groups

England 19.0

Native Born White Draft 13.2

Scotland 13.1

Holland 12.4

Total White Draft 12.0

Canada 11.1

Germany 10.1

Denmark 7.0

Sweden 5.9

Norway 5.3

Foreign Born White Draft 4.6

Ireland 4.3

Austria 4.1

Turkey 4.0

Russia 3.3

Belgium 2.9

Greece 2.2

Italy 1.5

Colored Draft 1.4

Poland 1.1

AMERICAN INTELLIGENCE 145

of feeble-mindedness in each group. The selection by the
draft boards probably excluded all idiots and many im-
beciles. The diagnosis of the border-line cases of feeble-
mindedness is, in the last analysis, a social diagnosis, and
can not be based on intelligence tests alone. It has been
found, however, that a "mental age" of eight indicates an
order of intelligence so low that the individual has diffi-
culty in adjusting himself to his environment adequately.
Table 32 gives a conservative estimate of the per cent, in
each nativity group below an approximate "mental age" of
eight. The percentages in Table 32 are shown graphically
in Figure 40.

Table No. 29
Per cent, of each nativity group in the D, D , and E

groups

England 8.8

Holland 12.0

Scotland 13.5

Germany 16.2

Denmark 17.0

Native Born White Draft 21.0

Canada 21.6

Sweden 23.2

Total White Draft 24.1

Norway 29.0

Belgium 29.3

Ireland 38.1

Austria 38.4

Turkey 43.6

Greece 44.6

Foreign Born White Draft 44.6

Russia 55.7

Italy 60.5

Poland 63.8

Colored Draft 67.5

ENGLAND

V//////////////////////////////////^^^^^

HOLLAND

Y /////////////////////////////////////////////^^^^

SCOTLAND

yy////////////////////////////////////^^^^^ -

GERMANY

y///////////////////////////////^^^^^

DENMARK

y// / //////////////////////////////^^^^^ z

NATIVE BORN WHITE DRAFT

y//////////////////////////////////////////////^^^^

CANADA

V/////////////////////////////////////////^^^^

SWEDEN

V////////////////////////////////////^^^^

TOTAL WHITE DRAFT

NORWAY

V//////////////////////////////////////////////y^
BELGIUM

W/////////////////////////////////////^^ I

IRELAND

y/////////////////////////////////////^^^^ I

AUSTRIA

VTTPTZ//////////////////// ////////^^^^

TURKEY

W7/////////////////////////////////?7r\

GREECE

V777P77/////////////////////////'/////A I

FOREIGN BORN WHITE DRAFT

MM II V//////////y/////A I

RUSSIA

ITALY

POLAND

V/////////// / /////////////A\
Wy///////////////////////A I
I II

NEGRO DRAFT

20 30 40 50 60

146

100

A&B

Figure 37. The relative standing of the various nativity groups in
the proportions of A and B men^ and D, D , and E men. In
interpreting this charts it should be remembered that A and B^
and D, D , and E do not represent absolute intelligence stand-
ards, but rather standards arbitrarily fixed. In this case the
standards were fixed by the 93,955 individuals making up the
sample of the total white draft, A and B representing scores
obtained by the upper 12% of this group, while D, D , and E
represent scores obtained by the lower 24%. The comparison is
relative and not altogether reliable, for it fails to take into con-
sideration the average, the number of cases, and the variability.
Tables 28, 29, and this chart have been presented for the
convenience of the reader, and to supplant Table 217, and Fig-
ure 19, on pages 697 and 698 of Memoir XV, which are not
based on combined scale results.

147

Table No. 30

Per cent, of each nativity group at or above the average
of the white officers

England 6.2

Scotland 4.8

Native Born White Draft 4.6

Holland. 4.3

Total White Draft 4.1

Canada 3.5

Germany 3.4

Austria 1.5

Sweden 1.4

Foreign Born White Draft 1.3

Ireland 1.2

Denmark 1.1

Norway 1.0

Turkey 0.8

Russia 0.8

Belgium 0.3

Italy 0.3

Greece 0.3

Poland 0.1

Figure 38. The proportion of each nativity group obtaining intel-
ligence scores at or above the average of the white officers
(18.84). Reference to Figure 31 will show that this criterion
indicates a relatively high order of intelligence. In comparing
this Figure with Figures 37 and 39, it should be noted that each
of the three figures has been drawn to a different scale. Our in-
terpretation of these figures must be made with caution, for we
are comparing extremes of the distribution curves without ref-
erence to the position of the average, the variability about the
average, or the number of cases in the various groups.

148

07o 1% 2% 37o 4% 5% 67o 7%

ENGLAND

SCOTLAND

NATIVE BORN WHITE DRAFT

HOLLAND

TOTAL WHITE DRAFT

CANADA

GERMANY

AUSTRIA

SWEDEN

FOREIGN BORN WHITE DRAFT

IRELAND

DENMARK

NORWAY

TURKEY

RUSSIA

BELGIUM

ITALY

GREECE

POLAND

0% l7o 2%

*% 5% Qfo 1%

149

Table No. 31

Per cent, of each nativity group at or below the average
of the negro draft

England 4.3

Holland 4.9

Germany 6.5

Scotland 6.8

Denmark. '. 7.5

Native Born White Draft. . 7.6

Sweden 11.5

Canada 11.6

Total White Draft 13.7

Norway 15.2

Belgium 16.0

Ireland 22.8

Austria 24.5

Greece 27.1

Turkey 28.2

Foreign Born White Draft 29.5

Russia 39.0

Italy 42.3

Poland 46.0

Figure 39. The proportion of each nativity group at or below the
average of the negro draft. Reference to Figure 31 will show
that this criterion indicates a rather low order of intelligence.
If 50% of any nativity group were at or below the negro aver-
age, the two distributions would be approximately identical.
Russia shows 39% below the negro average, Italy 42.3%, and
Poland 46%.

150

07o

109& 209& 80% 40% 50^

0% \07o 207e 30% 40%- 50%

151

Table No. 32

Per cent, of each nativity group below the approximate
"mental age" of eight

Holland 0.1

Germany 0.2

Denmark 0.2

England 0.3

Scotland. . : 0.4

Sweden 0.4

Norway 0.8

Canada 0.9

Native Born White Draft 1.1

Total White Draft 1.4

Belgium 1.6

Ireland 1.9

Greece 2.3

Austria 2.7

Turkey 3.1

Foreign Born White Draft 3.2

Russia 5.0

Italy 5.2

Poland 6.8

Colored Draft 10.0

Figure 40. Proportion of each nativity group testing below the ap-
proximate "mental age" of eight. This criterion indicates intelli-
gence of a very low order. These individuals are probably capa-
ble of adjusting themselves only to the simplest form of environ-
ment, occupation and conditions of living. Few of them would be
able to manage their affairs with ordinary prudence. Many of
them should be in custodial institutions.

159

POLAKD
NEGRO DRAFT

"1% 2% ^ *%> ^ 6% 7% 8% fo 10% n%

1.W

SECTION VII
RELIABILITY OF THE RESULTS

The results of the army psychological examination fig-
ured by means of the combined scale give us the best avail-
able measures of the intelligence of the various groups ex-
amined. Do these results apply to the army as a whole?
The logic underlying the answer to this question is the same
as that underlying the judgment of the whole by one of
its parts. The tea taster samples the tea to be graded. He
does not need to brew a whole bale of tea to find its worth.
In this experiment we have sampled the entire army by
taking 15,543 white oflScers, 93,955 white recruits and
23,596 negro recruits. Our group of white recruits was sub-
divided into 81,465 native born and 12,492 foreign born.

No one would hesitate to accept the results of the
81,465 native born as typical of the army as a whole. If we
continued sampling indefinitely, our results would increase
in reliability only as the square root of the number of
cases, and 81,465 cases constitute a sample that is a luxury
from the point of view of size. In the same way, no one
could seriously question the reliability of our sampling of
15,543 oflScers, 23,596 negroes and 12,492 foreign born as
typical of the remainder of the oflScers, negroes, and foreign
born whites in the army.

The results from the 81,465 cases in the native born
white draft may be taken as typical of white males between
the ages of 21 and 31 and above the idiot or imbecile grade.
In making our comparisons between other groups, we
know that the Selective Service Act called all men to the
colors impartially. The same regulations drew the Italians,
the negroes, the native whites, the Polish, and all other

154

AMERICAN INTELLIGENCE 155

groups into the army. The method of sampling all the
psychological records again drew these cases impartially.
If our theory of sampling is correct, then we may accept
the army results as very approximately typical of the male
population as a whole.

For instance, our figures in Table 3 show characteristic
differences in the average score on the army tests of foreign
born individuals in this country from to 5 yrs., and those
in this country from 6 to 10 yrs., etc. The same factors
which determined the sampling of the 3,576 cases in this
country to 5 yrs. determined the sampling of the 4,287
cases in this country 6 to 10 yrs. As long as the principles
of sampling are the same, we may take our small sample as
typical of the group as a whole.

The results of the psychological tests of foreign born
individuals classified according to length of residence,
taken as typical of our foreign born population as a whole,
indicate definitely that the average intelligence of succeed-
ing waves of immigration has become progressively lower.
Immigrants coming to this country between 1913 and 1917
have a lower average intelligence than those coming to
this country in the years 1908 to 1912. The group coming
to this country in the years 1903 to 1907 had a higher
average intelhgence that the 1908 to 1912 group, and a
lower average intelligence than immigrants coming to this
country in the years 1898 to 1902.

In drawing these conclusions we are taking the groups
examined in the army as typical of the corresponding
groups in the entire population. During the years 1913 to
1917, about 3 1/3 millons of immigrants came to this
country. We are actually using 3,576 cases or about 0.1%
as typical of the whole group. It may very properly be
objected that this is too small a sample on which to base
definite conclusions. We must therefore state our conclu-

156 AMERICAN INTELLIGENCE

sions less dogmatically, and with the proviso that if the
groups examined in the army are typical of the immigra-
tion coming to this country in the same periods, then we
know that our more recent periods of immigration give
us an average intelligence which becomes progressively
lower and lower. This tentative conclusion will be modified
by any evidence which tends to support the hypothesis
made.

The same kind of argument from the sample to the group
holds in our interpretation of the differences in the average
intelligence scores of groups in our army born in different
countries. For instance, in the period under consideration
from 1887 to 1917 there have been about 3 7/8 millions of
Italians, and over 3 million Russians who have come to
this country. We are actually using 4,009 Italians and
2,340 Russians as typical samples of these groups. Of
course no one would maintain that these 4,009 Italians are
typical of the population of Italy. There are so many vari-
able factors determining immigration that the immigrants
can not themselves be taken as representative of the
country as a whole. The question at issue is that of accept-
ing 4,009 Italians as typical of the 3 7/8 millions who have
come to this country since 1887. The chief claim to reli-
ability of our sample from each country is the fact that the
sample was drawn at random from the army group, and
the fact that the Selective Service Act drew the men from
each country impartially.

SECTION VIII
THE RACE HYPOTHESIS

The results of the examination of the nativity groups
suggest immediately that the race factor may underlie the
large differences found. If we do find the common factor of
race underlying the differences between the various nativ-
ity groups, it will give our results much greater reliability,
for the chance factors of sampling particularly inferior or
superior groups in the small nativity samples would dis-
appear in combination. Our figures are based on country
of birth and no statistics are available for race. The race
hypothesis must therefore be examined indirectly.

Writers on immigration, for the most part, divide the
countries of Europe into two groups (1) Northern and
Western, and (2) Eastern and Southern, and usually as-
sume that the immigration from Northern and Western
Europe has been mostly Nordic. This traditional method
is open to two very serious objections. In the first place,
the classification fails to differentiate the Alpine and Medi-
terranean race groups. In the second place the assumption
that the immigration from Northern and Western Europe
was mostly of a pure Nordic type is unwarranted, for this
classification includes Germany and Ireland, two countries
that have contributed very largely to our immigration in
the past. The following figures show the size of the Irish
and German immigration:

157

158

AMERICAN INTELLIGENCE

PER CENT.

TOTAL

FROM

DECADE

IMMIGRATION

FROM
IRELAND

FROM
GERMANY

IRELAND AND
GERMANY

1820-1830

143,439

35%

40%

1831-1840

599,125

35%

25%

71%

1841-1850

1,171,251

46%

1851-1860

2,598,214

35%

37%

72%

1801-1870

2,314,824

19%

34%

53%

1871-1880

2,812,191

15%

26%

41%

1881-1890

. 5,246,613

12%

28%

40%

1891-1900

3,844,420

11%

14%

25%

1901-1910

8,795,386

1911-1920

5,735,811

23^%

2>^%

These figures show clearly the fallacy of assuming that the
immigration from Northern and Western Europe has been
predominately Nordic, for Ireland is largely Mediterranean
and Germany largely Alpine.

If we wish to obtain even approximate estimates of the
contributions of each of the three European races to our
importations, it is necessary to abandon the Northern and
Western, and Eastern and Southern classification and try
another method. If it were possible to make even approxi-
mate estimates of the percentage of Nordic, Alpine and
Mediterranean blood in each of the European nations send-
ing immigrants to this country, such approximate estimates
would be very much superior to the present method.

In collaboration with students of this subject, I have
constructed Table 33 which contains tentative estimates of
the present blood constitution of the countries sending im-
migrants to this country. This table is, of course, only an
approximation to the truth, and many persons will dis-
agree with the estimates. For this reason, I am re-publishing
in Table 34, Table 68, page 100, of the Statistical Abstract
for the United States for 1920, which shows the arrivals of
alien passengers and immigrants by nationalities and by

Table No. 33

Tentative estimates of the proportion of Nordic, Alpine
and Mediterranean blood in each of the European coun-
tries.

Austria-Hungary

Belgium

Denmark

France

Germany

Greece

Italy

Netherlands

Norway

Sweden

Russia (including Poland)

Poland

Spain

Portugal

Roumania

Switzerland

Turkey (unclassified)

Turkey (in Europe) (including Serbia,

Montenegro and Bulgaria)

Turkey (in Asia)

England

Ireland

Scotland

Wales

British North America

EB CENT.

PER CENT.

NORDIC

ALPINE

MEDITERRANEAN

100

159

Table No. 34

No. 68 ARRIVALS OF ALIEN PASSENGERS AND IMMIGRANTS, 1820 TO 1920: Bt Nationau-

TIE8 AND BT DeCADES.

[Sources: Records of the Bureau of Statistics prior to 1896; subsequently, reports of the Commissioner General
of Immigration, Department of Labor. The figures represent "alien passengers" from Oct. 1, 1820, to Dec. 31,
1867; "immdgrants" from Jan. 1, 1868, to date.

COUNTBY OF LAST PERMANENT
RESIDENCE

OCT. 1, 1820,
TO SEPT. 30, 1830

OCT. 1, 1830,
TO DEC. 31, 1840

JAN. 1, 1841,
TO DEC. 31, 1850

JAN. 1, 1851,
TO DEC. 31, 1860.

169

8,497

6,761

1,063

45,575

152,454

5,074

639

77,262

434,626

4,738

Denmark

3,749

76,358

Germany

951,667

Greece 1 . ...

Italy

408
1,078

}

91
} 2,622

2,253
1,412

1,201

646

2,954

1,870
8,251

13,903

656

2,759

9,231

Netherlands

10,789

Norway

20,931

Russia, including Russian Poland^

Spain^

1,621

Portugal*

10,353

Switzerland

3,226

4.821

4,644

25,011

T'lirtpv in RiirnnpS

United Kingdom:

England

22,167

2,912

50,724

73,143

2,667

207,381

263,332

3,712

780,719

385,643

Scotland

38,331

914,119

Wales*

Total United Kingdom

75,803

283,191

1,047,763

1,338,093

155

116

Total Europe

98,816

495,688

1,597,502

2,452,657

British North America^

2.277
4,817

105
3,834

531

13,624

6,599

12,301

856

41,723

3,271

368

13,528
3,579

59,309

Mexico^

3,078

449

West Indies: Bermuda and Miquelon.

10,660

1,224

Total America^

11,564

33,424

62,469

74,720

Islands of the Atlantic

352

103

337

3,090

China

41,397

India^

Other Asia

Total Asia

41,458

32,679

69,801

29
55

52,777

158

Total Africa

210

All other countries

25,921

Grand total

143,439

699,125

1,713,251

2,598,214

ilncluded in "Europe, not specified," prior to 1891-1900. ^Includes also Finland after 1872.
Includes Canary and Balearic Islands after 1900. , . , i * .

Figures include the Azores and Cape Verde Islands after 1879, they being classed with Portugal so far as that
country is separately shown.

160

JAN. 1, 1861,

TEARS ENDED JUNE 30

TO JUNE 30, 1870

1871 TO 1880

1881 TO 1890

1891 TO 1900

1901 TO 1910

1911 TO 1920

7.800

6,734

17,094

85,984

787,468

72,969

7,221

31,771

72,206

718,182

353,719

20,177

88,132

50,464

1,452,970

597,047

20,062

52,670

36,006

543,922

15,996

655,694

31,816

( 95,265

\ 230,679

693,703

( 6,723

( 23,010

14.559

33.149

2.562

2,145,266

41,635

65,285

73,379

341,498

167,519

2,045,877

48,262

190,505

249,534

1.597,306

27,935

69,149

53,008

34,922

118,202

896,342

33,746

41,983

61,897

143,945

184,201

11,728
9,102

109,298
4,536
8,493

55,759
16,541

211,245

52,254

9,893

307,309
53,701

568,362

265,088

6,535

1,109,524
43,718
66,395
95,074
921,957
68,611
89,732
13,311

23,286

28,293

81,988

23,091
77,098

668,128

38,768

435,778

460.479

87,564

436,871

657,488
149,869
655,482

271,094
60,053

403,496
11,186

388,017

120,469

339,065

17,464

249,944
78,601

145,937
13,107

1,042,674

984.914

1,462,839

745,829

865,015

487,589

210

656

10,318

4,370

1,719

18,350

2,064,407

2.261,904

4,721,602

3,703,061

8,136,016

4,376.564

153,871

2,191

9,043

1,396

383.269

5,362

210

13,957

928

392,802

1,913

462

29,042

2,304

2,631

746

1,183

35,040

3,059

179,226

49,642

8,112

107,548
17,280

742,185
219,004

17,159
123,424

41,899

166,597

403,726

426,523

42,659

361,808

1,143,671

3,446

10,056

15,798

64,301

123,201

61,711

23,166

26,855

8,398

28,370

20,605

4,713

129,797

77,393

11,059

21,278
2,082

83,837

79,389

308

622

6,669

5,973

64,609

123.823

68,380

86,815

243,567

192,559

221

312

15,232

10,913

229

1,540

12,574

437

1,299

8,793
1,343
1,749

12,973

7,368

33,654

13,427
8,443
1,147

2,314,824

2,812,191

5,246,613

3,844,420

8,795,386

6,735.811

^Includes Serbia. Bulgaria, and Montenegro prior to 1920; included in "Europe, not specified," prior to 1891-
1900; also, after 1919, Czechoslovakia, Poland, and the Kingdom of the Serbs, Croats, and Slovenes.
Not separately stated prior to 1891-1900.

'Immigrants from British North America and Mexico were not reported from 1886 to 1895, inclusive.
8Not separately enumerated prior to 1899.

161

162 AMERICAN INTELLIGENCE

decades from 1820 to 1920. My own Tables 9 and 10 give
the distribution of the intelhgence scores on the combined
scale for the nativity groups we are studying. Anybody
who disagrees with the estimates given in Table 33 may
take these tables and split them according to any other
estimates he wishes to make. However, minor changes in
the proportions given in Table 33 would make very little dif-
ference in, the final results. The figures which follow are
merely estimates based on Table 33. I am not claiming
that these figures are absolutely reliable, but merely that
they represent very much closer approximations to the
truth than would be obtained from the Northern and West-
ern, and Southern and Eastern classification.

To obtain an estimate of the proportion of Nordic, Al-
pine, and Mediterranean blood in our immigration since
1840, the immigration figures by countries, given in Table
34, have been cut according to the proportions given in
Table 33 and re-combined into percentage estimates which
are given in Table 35. These estimates show in general an
immigration prior to 1890 which ran 40% or 50% Nordic
blood.

Since 1890, the proportion of Nordic blood has dropped
to 20% or 25%, the Alpine stock now constituting about
50% of the total and the Mediterranean 20% or 25%.

The proportions given in Table 35 are shown graphically
in Figure 41. The percentage estimates, given in Figure 35
and shown graphically in Figure 41, should be considered
in connection with the total volume of immigration for
each decade given in Table 34 and shown graphically in
Figure 42.

Table No. 35

Estimate of the amount of Nordic, Alpine and Mediter-
ranean blood coming to this country from Europe in each
decade since 1840.

PEB CENT.

PER CENT

PER CENT.

TOTAL

NORDIC

ALPINE

MEDITERRANEAN

OTHERS AND

DECADE

IMMIGRATION

BLOOD

UNCLASSIFIED

1841-1850

1,713,251

40.5

19.0

36.2

4.3

1851-1860

2,598,214

42.3

25.5

28.9

S.3

1861-1870

2,314,824

50.6

26.0

19.2

4.2

1871-1880

2,812,191

48.8

28.5

16.7

6.0

1881-1890

5,246,613

46.1

85.2

16.0

2.7

1891-1900

3,844,420

30.2

43.8

22.5

8.5

1901-1910

8,795,386

19.8

51.3

24.3

4.6

1911-1920

5,735,811

22.6

44.0

23.7

9.7

163

1 s

^ I

W '

1 o

W '

,g'

go ^^13^^

^ en 43 G^_fl ^

o a, ^ N '^ ^ ^

S ^ ^ vQ O (u
1? > X

.t^ C 550 3 (^ - -g 0^

^ C 3 W

03 a a g

M ^! '^ "73 -^ ^ '*^

SV, f:2 a fl <i^ -=? "
-^ a; 9 o .2

O ^ c3 O CJ >H
. - ^ =3

^ ^ '3 -^ -^ - H

:3 O fl >H rj O
^' ^ 5yo 0, - .S o

C3 C S3

> "Jj O -N >^

o a^

w S e_

>> O

03 a;

.2 o "^iij M

CO ^j

<U CO >!

H a _ri ,_( O ' I

oc > <u a g

b3 _^ > O c3 ^

o G

a ::

CO
CO

^ ^ .3^ 1 S p:5

-^ '-^ ^

. 2 o S>~ C

553 ^ r^ -H ^ a

cj ^ <i; ^ o '^

-^ rrt -^ T3 ^

.2 n3 fl w a <

-M o c3 3 ^
C3 ^ ^ > ^

^^ S o o o

2 a T3 00 '-' ?

-' <^ 3 o W O

^ ;3 c ^ .9 o

05 ^pq C5

. ^ S ^ O T! .

-~ w (U

s ;r -^ -^

168 AMERICAN INTELLIGENCE

In order to obtain an estimate of the intelligence of the
three European races in this country, the distributions of
the intelligence scores on the combined scale given in
Table 9 were cut according to the proportions given in
Table 33, and re-combined into Nordic, Alpine, and Medi-
terranean groups. The final distributions are, of course,
neither purely Nordic, Alpine, nor Mediterranean, but the
sample of individuals we have thus selected as Nordic is
undoubtedly more typical of the Nordic race type than
it is of the Alpine and Mediterranean types. In the same
way, the Alpine and Mediterranean groups are more typi-
cal of each of these race types than they are of either of the
other two. With thus much of apology for the method, I
will, in the following pages, simply for brevity of expres-
sion, call these groups Nordic, Alpine, and Mediterranean.
The reader must bear in mind that the distributions are
only approximate samplings.

The actual distributions on the combined scale of the
three race groups so selected are given in Table 36, togeth-
er with the proportions in each thousand. The distribution
curves of the three groups are shown in Figure 43, in which
the horizontal direction represents scores on the com-
bined scale, and the vertical direction proportions in each
thousand making each intelligence score.

The differences found are very marked. The difference
between the Nordic and Alpine group is 1.61 =1=0.042, a
difference which is 38.3 times the probable error of the
difference. The difference between the Nordic and Medi-
terranean group is 1 .85 =*= 0.042, a difference which is 44 times
the probable error of the difference. The Alpine and Medi-
terranean groups are, on the other hand, very much closer
together, the difference being 0.24 =*= 0.04, a difference which
is 6 times the probable error of the difference.

The easiest and most obvious objection that can be made

Table No. 36

Analysis of the foreign born white draft by races. Distri-
butions of the inteUigence scores of the Nordic, Alpine
and Mediterranean groups.

COMBINED

ACTUAL DISTRIBUTION

PROPORTION IN EACH

SCALE

THOSUAND

INTERVALS

NORDIC

ALPINE

MEDITER-
RANEAN

NORDIC

ALPINE

MEDITER-
RANEAN

24.0-24.9

23.0-23.9

....

22.0-22.9

....

21.0-21.9

20.0-20.9

"'2

19.0-19.9

18.0-18.9

17.0-17.9

135

16.0-10.9

238

155

103

15.0-15.9

357

246

180

103

14.0-14.9

469

372

296

136

13.0-13.9

566

544

408

164

114

111

12.0-12.9

528

650

591

153

136

141

11.0-11.9

371

628

590

107

132

140

10.0-10.9

260

595

509

125

136

9.0- 9.9

184

546

523

115

125

8.0- 8.9

112

403

376

7.0- 7.9

248

223

6.0- 6,9

124

108

5.0- 5.9

4.0- 4.9

3.0- 3.9

....

2.0- 2.9

....

1.0- 1.9

....

No. of cases.

3456

4766

4196

Average ....

13.28

11.67

11.43

S.D

2.70

2.87

2.70

d 3i S

s 2 o 2

w ^ -S^ ^ 5^ g

5 ;^ "^ *^. ^

O "P ^ CO <u

*^ *H 0) a; <u

V "^ <^ 5 ^

I- ^2:^ ?

O H W .

C 03 ^ d <^
DC o (U OJ >^

O e3 : 0) -^

C a O c3 o

'C -^ ^ ^

^ o ^

"^ ^g MH <D ^

^^^ a;

AMERICAN INTELLIGENCE 171

to these findings is that the superiority of the Nordic group
is due to the fact that it contains so many EngUsh speaking
persons, and that lack of facihty in the use of EngHsh is a
handicap to the non-Enghsh speaking foreign born in the
army tests. We have previously examined this hypothesis
in connection with the argument establishing the fact that
each succeeding five year period since 1902 shows a gradual
deterioration in the intelligence of the immigrants examined
in the army, and have definitely shown that the language
factor does not distort the scores of the years of residence
groups. There is, however, a considerable amount of wish-
ful thinking on the subject of race, and it is well to make
assurance doubly sure by testing the hypothesis that the
superiority of the Nordic group is caused by the presence
in the group of English speaking populations.

It is possible to split the Nordic distribution in such a
way that one group will contain representatives from
countries which are predominantly English speaking (Eng-
land, Scotland, Ireland and Canada), while the other group
will contain representatives from countries which are pre-
dominantly non-English speaking (Holland, Denmark,
Germany, Sweden, Norway, Belgium, Austria, Russia,
Italy and Poland) . This we have done, and the results are
given in Table 37, the two distributions being shown in
Figure 44.

The distributions of the English speaking Nordic group
and the non-English speaking Nordic group show a differ-
ence of 0.87=1=0.065, a difference which is 13.4 times the
probable error of the difference. There are, of course, cogent
historical and sociological reasons accounting for the in-
feriority of the non-English speaking Nordic group. On the
other hand, if one wishes to deny, in the teeth of the facts,
the superiority of the Nordic race on the ground that the
language factor mysteriously aids this group when tested,

Table No. 37

Analysis of the total Nordic sample into an English speak-
ing Nordic group and a non-English speaking Nordic
group.

COMBINED

ACTUAL DISTRIBUTION

PROPORTION IN EACH

SCALE

THOUSAND

INTEKVAT,a

ENGLISH

NON-ENGLISH

ENGLISH

NON-ENGLISH

SPEAKING

NORDIC

24.0-24.9

....

23.0-23.9

22.0-22.9

....

21.0-21.9

' k

20.0-20.9

19.0-19.9

18.0-18.9

17.0-17.9

16.0-16.9

108

181

15.0-15.9

143

214

116

14.0-14.9

176

298

148

132

13.0-13.9

201

865

168

164

12.0-12.9

172

856

189

160

11.0-11.9

109

262

118

10.0-10.9

189

9.0- 9.9

185

8.0- 8.9

7.0- 7.9

6.0- 6.9

5.0- 5.9

4.0- 4.9

....

3.0- 3.9

....

2.0- 2.9

....

1.0- 1.9

....

No. of cases. . .

1234

2222

Average

13.84

12.97

S.D

2.79

2.60

173

O <L> jQ pfi .JH

be ^ O

52 ^-Sst!

2 .2 o bcna

^ i3!tj ^ t:

^ E2 :i b^ S S

JZ! i-M 'ft

si " ""I

ej .rt ^_> -^

1 O O g ^ C

O O bC-^ g^

.^^ g S. 2 .3^

.s :3 .S^ ;S ^ ^

'^ 'IS . .S ^ V5

d 'bc5 *<-i

';P go

rt .S ^ ;=3 H c

- _r .-;:; fl" C

QJ "^ i-H jj ~

-< -' tJ r!

tn < Ji o

{>..S:J O

S fl a.

^ p g .^:S S S^
V. o .S ^ 9^ o

- \g So

S JH 03
.Oi bC^

174 AMERICAN INTELLIGENCE

he may cut out of the Nordic distribution the English
speaking Nordics, and still find a marked superiority of the
non-English speaking Nordics over the Alpine and Medi-
terranean groups. The difference between the non-English
speaking Nordic group and the Alpine group is 1.30 =*= 0.047,
a difference which is 27.6 times the probable error of the
difference. The difference between the non-English speak-
ing Nordic group and the Mediterranean group is 1.54=1=
0.047, a difference which is 31.3 times the probable error
of the difference. The distributions are shown graphically
in Figure 45. Discarding the English speaking Nordics
entirely, we still find tremendous differences between the
non-English speaking Nordic group and the Alpine and
Mediterranean groups, a fact which clearly indicates that
the underlying cause of the nativity differences we have
shown is race, and not language.

It may be convenient for some to interpret the differ-
ences found between the representatives of the three Euro-
pean races in this country in terms of the standards having
popular significance which were used in Section VI. The cri-
teria of the per cent. A and B, and the per cent. D, D
and E give the following results:

PER CENT. PER CENT.

A AND B D,D AND E

English speaking Nordic 12 . 3 19.9

Total Nordic 8.1 25.8

Non-English speaking Nordic . 5.7 29 . 1

Alpine 3.8 50.3

Mediterranean 2.5 53 . 6

The criteria of the per cent, at or above the average
white officer, and at or below the average of the negro
draft give the following results:

S3 t^

-^ ^^

C -M

o o

bf) 'J5 a C O rQ c >j

Q, a> <u ;:5 "^ -^ ^^ ^

'^^ fl '^ _- -^-' Jzj
<v .ti -^f^ Ti -^ *'

Ph.22 "*^ ea "1^ o
c

5 "XJ . >^ C oj a
2 ^ d .t^ ^ C C

g ;^. ^ .2 P -^^g^
g bc rt S 2 *^ ^
c .S ^ &- ^ ^

o ^ ;:5 bc q ?J 0)
^W I i S " c
-M G o 2; = o fl

S -p. ^ ^ ^ ^ 2

^ < ^ -B '? ^ a ^

bo p.^ S c' o o

I H j^ M M

^ .2

be Sf

C3 rO

o o

bC'-S

o ^

176 AMERICAN INTELLIGENCE

PER CENT. PER CENT.

AT OR ABOVE AT OR BELOW

AVERAGE AVERAGE OF

WHITE THE NEGRO
OFFICER DRAFT

English speaking Nordic 4.0 10 . 9

Total Nordic 2.3 14.5

Non-English speaking Nordic . 1.3 16.5

Alpine 1.0 34.5

Mediterranean 0.5 36 . 5

The criterion of the per cent, below an approximate
**mental age" of eight gives the following results:

PER CENT.
BELOW

"mental
age" 8

English speaking Nordic 0.8

Total Nordic 1.1

Non-English speaking Nordic . 1.3

Alpine 4.2

Mediterranean 4.2

SECTION IX

RE-EXAMINATION OF PREVIOUS CONCLUSIONS
IN THE LIGHT OF THE RACE HYPOTHESIS

It is now necessary to retrace our steps for a moment to
examine some of our previous conclusions in the light of
this new hypothesis. The hypothesis that the differences
between the nativity groups found in the army tests are
due to the race factor may be used to re-test our previous
conclusions that each succeeding five year period of immi-
gration since 1902 has given us an increasingly inferior
selection of individuals (Section IV) . The periods which we
sample by means of the army data, and the average score
on the combined scale of each sample are as follows :

PERIOD

NUMBER OF
CASES

COMBINED

SCALE
AVERAGE

1887-1897

764

13.82

1898-1902

771

13.55

1903-1907

1897

12.47

1908-1912

4287

11.74

1913-1917

3576

11.41

Table 35, which gives our estimates of the per cent, of
Nordic, Alpine and Mediterranean blood coming to this
country, shows that the big change in immigration came
between the decades 1881-1890 and 1891-1900, the per-
centage of Nordic blood which formerly ran from 40% to
50% having dropped to 30% in the decade 1891-1900, and
to approximately 20% or 25% in the two subsequent dec-
ades. On the other hand, the big drop in the intelligence
of immigrants arriving came after 1902. The change in

177

178 AMERICAN INTELLIGENCE

character of the immigration would account for part of the
decHne in the average intelligence of succeeding periods of
immigration, but not for all of it. The decline in intelligence
is due to two factors, the change in the races migrating to
this country, and to the additional factor of the sending
of lower and lower representatives of each race.

The only tendency which would relieve this deplorable
situation would be a current of emigration strong enough
to counteract the current of immigration. Table 6 preced-
ing shows the ratio between emigration and immigration
for each of the nativity groups involved in this study, and
we find in general between 1908 and 1917 a return current
approximately one third of the arriving current.

Unfortunately, no emigration statistics are available
prior to 1908, and the figures after 1912 are distorted by
the Balkan and European wars. The only sample that we
can take that is comparatively free from outside influences
is the sample 1908-1912. Taking the figures of arrivals and
departures for this period, and dividing them into Nordic,
Alpine and Mediterranean groups according to the method
previously outlined, we obtain the following percentage
estimates :

ALIEN

TMMTGRANTS
ADMITTED

EMIGRANTS
DEPARTED

NET
IMMIGRATION

Per cent, of Nordic

blood

21.2

16.0

23.9

Per cent, of Alpine

blood

50.4

50.6

50.2

Per cent, of Mediter-

ranean blood

23.2

28.6

20.5

Per cent, others and

unclassified

5.2

4.8

5.4

AMERICAN INTELLIGENCE 179

The sample from this five year period shows a shght
change (approximately 3%) in favor of the Nordic type
and against the Mediterranean type, the Alpine immigra-
tion holding its own. There is therefore no relief from our
receding curve of intelHgence from emigration, if this five
year period be taken as typical of the outward alien pas-
senger movement in other years.

I?- It will be remembered that the army authors tentatively
offered the hypothesis that the more intelhgent immigrants
remained in this country, while the more stupid ones went
home, as a possible method of accounting for the increase
of intelhgence scores with increasing years of residence.
The gain of 3% in favor of the Nordic immigration would
produce a very slight tendency in this direction, but not
enough to account for the actual increase of intelligence
scores found with increasing years of residence, 11.41
(1913-1917) to 13.82 (1887-1897).

It will also be remembered that the army writers offered
the hypothesis of the better adaptation of the more thor-
oughly Americanized group to the situation of the examin-
ation to account for the increases shown. The factor of the
adaptation to the situation of the examination cannot be
dissected out of the total scores of the test. If such a factor
were present, it would fall equally heavity on Nordic, Alpine
and Mediterranean alike, unless the change in the character
of immigration were so complete that the groups sampled
at the two extremes of the residence groups (1887-1897
and 1913-1917) represented different race groups.

But the difference between these two years of residence
groups (2.41 =t 0.0735) is so marked that it would be neces-
sary to assume (if our Nordic group were the more thor-
oughly Americanized) that the 1887-1897 group was com-
posed entirely of English speaking Nordics or their equiva-
lent in intelligence, and that our 1913-1917 group was

180 AMERICAN INTELLIGENCE

composed entirely of Mediterraneans or their equivalent
in intelligence, assumptions quite unwarranted in view
of the fact that in the two years of residence groups 1887-
1897 and 1898-1902 we sampled 1545 individuals, while
our Nordic group includes 3456 cases, and also in view of
the fact that the Nordic immigration has dropped, in the
period observed, at the outside from 45% to 20%. We may
therefore conclude that the intangible factor of "the more
thoroughly Americanized group" can not be used to ex-
plain the high test record of the Nordic group.

There is only one other possible escape from the con-
clusion that our test results indicate a genuine intellectual
superiority of the Nordic group over the Alpine and Medi-
terranean groups, and that is the assumption that the
situation of the examination involved a situation that was
"typically Nordic." This assumption of course lands us in
a perfect circle of reasoning. It would leave us with the
conclusion that there was something mysteriously Nordic
about alpha and beta that favored this race. We should
have to assume that the Nordic, no matter where he may
be, in the Canadian Northwest, in the Highlands of Scot-
land, or on the shores of the Baltic, is always ready for an
intelligence test. Perhaps it would be easier to say that the
Nordic is intelligent. A situation "typically Nordic" could
not be used, however, to account for the slight but real dif-
ference between the English speaking Nordic and the non-
English speaking Nordic groups. It is therefore best to
abandon the attempt to account for the differences by the
more or less feeble hypotheses that would make these dif-
ferences an artifact of the method of examining, and recog-
nize the fact that we are dealing with real differences in the
intelligence of immigrants coming to our shores.

We have previously noted the fact that the foreign born
in the army sampled as representative of the immigrants

AMERICAN INTELLIGENCE 181

coming to this country between 1887 and 1897 were statis-
tically identical with the native born white draft. The
change in the character of our immigration came between
1890 and 1900. The real drop in the curve of intelligence,
however, started about 1900. We, therefore, cannot account
for the drop in the intelligence of the immigrants sampled
as representatives of those coming to this country in each
five year period since 1902 by the race hypothesis entirely.

SECTION X

COMPARISON OF OUR RESULTS WITH THE

CONCLUSIONS OF OTHER WRITERS

ON THE SUBJECT

In a very definite way, the results which we obtain by
interpreting the army data by means of the race hypothesis
support Mr. Madison Grant's i thesis of the superiority of
the Nordic type: "The Nordics are, all over the world, a
race of soldiers, sailors, adventurers, and explorers, but
above all, of rulers, organizers, and aristocrats in sharp
contrast to the essentially peasant and democratic char-
acter of the Alpines. The Nordic race is domineering, in-
dividualistic, self-reliant, and jealous of their personal
freedom both in political and religious systems, and as a
result they are usually Protestants. Chivalry and knight-
hood and their still surviving but greatly impaired counter-
parts are peculiarly Nordic traits, and feudalism, class
distinctions, and race pride among Europeans are traceable
for the most part to the north." (p. 228.) "The pure Nordic
peoples are characterized by a greater stability and steadi-
ness than are mixed peoples such as the Irish, the ancient
Gauls, and the Athenians, among all of whom the lack of
these qualities was balanced by a correspondingly greater
versatility." (pp. 228-229.)

Our results based on the army data also support Mr.
Grant's estimates of the Alpine race: "The Alpine race is
always and everywhere a race of peasants, an agricultural
and never a maritime race. In fact they only extend to
salt water at the head of the Adriatic and, like all purely

iMadison Graot. The Passing of the Great Race. New York, 1922, Pp. 476.

182

AMERICAN INTELLIGENCE 183

agricultural communities throughout Europe, tend toward
democracy, although they are submissive to authority both
political and religious, being usually Roman Catholics in
western Europe. This race is essentially of the soil, and
in towns the type is mediocre and bourgeois." (p. ^27.)

Our results also support de Lapouge^ in his contention
that the Nordic type is superior to the Alpine. He says con-
cerning the Alpine: **I1 est le parfait esclave, le serf ideal,
le sujet modele, et dans les republiques comme la notre,
le citoyen le mieux vu, car il tolere tons les abus." (p. 233.)
"Les etats brachycephales, France, Autriche, Turquie, sans
parler de la Pologne qui n'est plus, sont loin d'offrir la
vitalite des Etats -Unis ou de I'Angleterre. Cependant
la mediocrite meme du brachycephale est une force. Ce
neutre echappe a toutes les causes de destruction. Noiraud,
courtaud, lourdaud, le brachycephale regne aujourd'hui
del'Atlantique a la Mer Noire. Comme la mauvaise monnaie
chasse Tautre, sa race a supplante la race meilleure. II est
inerte, il est mediocre, mais se multiplie. Sa patience est
au-dessus des epreuves; il est sujet soumis, soldat passif,
fonctionnaire obeissant. II ne porte pas ombrage, il ne se
revolte point." (p. 481.)

It must, however, be frankly admitted that our results,
which show the Mediterranean race inferior to the Alpine,
are in contradiction with those of most writers who have
inferred the intellectual level of a race from its historical
achievements. Mr. Grant, for instance, says: "The mental
characteristics of the Mediterranean race are well known,
and this race, while inferior in bodily stamina to both the
Nordic and the Alpine, is probably the superior of both,
certainly of the Alpines, in intellectual attainments. In the
field of art its superiority to both the other European races

1 Georges Vacher de Lapouge. UAryen, son role social. Paris, 1899, Pp. 563.

184 AMERICAN INTELLIGENCE

is unquestioned, although in hterature and in scientific re-
search and discovery the Nordics far excel it." (p. 229). ^

The apparent contradiction between our results and the
estimates of other observers has a very obvious solution,
viz., that those who draw their conclusions from historical
data are studying the Mediterranean race as it was at the
period of its greatest development, when it produced the
civilizations of Egypt, Mesopotamia, Phoenicia, and Crete,
and, with a Nordic predominance, gave the civilizations
of Greece and Rome, while our data sample this race group
as it is at the present time.

The sample we have taken as representative of the Medi-
terranean race, as it is now constituted, is drawn from im-
migrants in our army born, for the most part, in Greece,
Ireland, Italy, and Turkey, and inasmuch as the number
from Italy (4009) is so large, our Mediterranean sample
is heavily weighted (approximately 2/3) by this nativity
group.

In regard to the Irish, Mr. Madison Grant says: "In
spite of the fact that Paleoliths have not been found there,
some indications of Paleolithic man appear in Ireland, both
as single characters and as individuals. Being, like Brittany
situated on the extreme western outposts of Eurasia, it has
more than its share of generalized and low types surviving
in the living populations, and these types, the Firbolgs,
have imparted a distinct and very undesirable aspect to a
large portion of the inhabitants of the west and south and
have greatly lowered the intellectual status of the popula-
tion as a whole. The cross between these elements and the

^The quotations I have chosen from Mr. Madison Grant's chapter on Racial
Aptitudes most certainly do not do justice to that author, but they seemed to
me to summarize his general position briefly. The entire book should be read to
appreciate the soundness of Mr. Grant's position and the compelling force of his
arguments.

AMERICAN INTELLIGENCE 185

Nordics appears to be a bad one, and the mental and cul-
tural traits of the aborigines have proved to be exceedingly
persistent and appear especially in the unstable tempera-
ment and the lack of coordinating and reasoning power,
so often found among the Irish. To the dominance of the
Mediterraneans mixed with Pre-Neolithic survivals in the
south and west are to be attributed the aloofness of the is-
land from the general trend of European civilization and
its long adherence to ancient forms of religion and even to
Pre-Christian superstitions." (pp. 202-203.)

The immigrants in this country from Italy come mostly
from Southern Italy and Sicily. The following quotation
from Ripley 1 concerning Sicily is significant:

"Commanding both straits at the waist of the Mediter-
ranean, it has been, as Freeman in his masterly description
puts it, 'the meeting place of the nations.' Tempting, there-
fore, and accessible, this island has been incessantly over-
run by invaders from all over Europe Sicani, Siculi, Feni-
cii, Greeks, and Romans, followed by Albanians, Vandals,
Goths, Saracens, Normans, and last by the French and
Spaniards. Is it any wonder that its people are less pure in
physical type than the Sardinians or even the Calabrians of
the mainland near by.^^ Especially is this noticeable on its
southern coasts, always more open to colonization than the
northern edge. Nor is it surprising, as Freeman rightly
adds, that 'for the very reason that Sicily has found dwell-
ing places for so many nations, a Sicilian nation there
never has been.'" (p. 271.)

The secret of the whole dilemma is the intermingling of
races around the Mediterranean littoral in the last 2500
years. It is beside the point to contrast our results obtained
by the actual psychological measurements of living repre-

iWilliam Z. Ripley. The Races of Europe. New York, 1899, Pp. 624.

186 AMERICAN INTELLIGENCE

sentatives of this race with the attainments of the tem-
porary civiHzations that flared up in historical times. The
whole question of the degeneration of these peoples has
been discussed by Mr. Charles W. Gould,^ and our results
from the examinations of drafted men born in these regions
support his position.

It is rather difficult to compare our results from the
race groups with the various hypotheses erected by Pro-
fessor William McDougall,^ who, while he does not claim
for the Nordic race "any general innate superiority" (p.
29), analyzes the mental constitution of this race and the
other European races in such a way that an examination
of his theories will be interesting. Professor McDougall's
hypotheses, very briefly and inadequately stated are: that
the Nordic is stronger in the instinct of curiosity, the root
of wonder, than the Mediterranean ; the herd instinct, the
root of sociability, is stronger in the Mediterranean than in
the Nordic; the Nordic is constitutionally introvert, the
Mediterranean constitutionally extrovert; the instinct of
self-assertion is strong in the Nordic; the Alpine is introvert
but not so strongly introvert as the Nordic; the Alpine has
a high degree of sociability, is perhaps relatively weak in
curiosity, and strong in the instinct of submission.

In discussing innate differences in instinctive endowment,
psychologists are still more or less in a speculative realm,
but the field is open to experimental attack, and a body of
knowledge based on experimentation is gradually growing.
At the present time we must rely on consensus of opinion
rather than experiment. A census of text-books on psychol-
ogy would show "curiosity" usually listed as an instinctive

iCharles W. Gould. America, a Family Matter. New York, 1922, Pp. 196.
^William McDougall. Is America Safe for Democracy. New York, 1921, Pp. 213.

AMERICAN INTELLIGENCE 187

tendency. If we follow Professor Thorndikei in his analysis,
and eliminate many tendencies that others include, we shall
still have left the instincts of multiform mental and phy-
sical activity as the potent movers of men's economic and
recreative activities, (p. 144.) If any instinctive tendency
finds expression in the tasks assigned by the army tests, it
is this instinct for multiform mental activity, more vaguely
termed "curiosity." Our tests, however, measure the end
result of such a tendency and not the tendency itself, and
it is only in this vague way that our results showing the
definite intellectual superiority of the Nordic race can be
taken as substantiating or contributing to Professor Mc-
Dougall's hypothesis.

It is difficult to check our results from the analysis of the
foreign born white draft by country of birth (reported in
Section VI) with the results of other investigators, on ac-
count of the different tests that were used, and the differ-
ent methods of selecting subjects. Miss Murdoch 2 exam-
ined, by means of the Pressey group point scale, 500 Jew-
ish children and 500 Italian children at one school in New
York City, and 500 American children and 230 negro
children at another school. The American and Jewish chil-
dren tested about the same. About 15% of the Italians
equalled or exceeded the median of the Jews, and about
30% of the negroes equalled or exceeded the median of the
Jews. The investigation equalizes the environmental factor
by selecting, in one instance, Italians and Jews from the
same school and consequently from the same general
neighborhood (East 110th St. near 2nd Ave.), and, in the
other instance, by selecting native white and negro chil-

lE. L. Thorndike, Educational Psychology^ Vol. I. The Original Nature of Man,
New York, 1919, Pp. 327.

2K. Miirdoch. A Study of Race Differences in New York City. School and Society,
1920, 11, pp. 147-150.

188 AMERICAN INTELLIGENCE

dren from the same general neighborhood (West side, 8th
Ave., near 140th St.), but the American children living in
this neighborhood can not be taken as typical of the
country as a whole.

Miss Arlitti concludes from her examination of 343 chil-
dren, (191 native born Americans, 87 Italians, and 71 ne-
groes), by the Stanford-Binet scale, that "there is a marked
difference in the distribution of intelligence in groups of
the same race but different social status," and states that
"race norms which do not take the social status factor into
account are apt to be to that extent invalid." (p. 183.)
This position seems to ignore the observation, repeatedly
confirmed by experiment, that children from the profes-
sional, semi-professional and higher business classes have,
on the whole, an hereditary endowment superior to that
of children from the semi-skilled and unskilled laboring
classes. Termans states "It has in fact been found wherever
comparisons have been made that children of superior
social status yield a higher average mental age than chil-
dren of the laboring classes. . . . However, the common
opinion that the child from a cultured home does better in
tests solely by reason of his superior home advantages is
an entirely gratuitous assumption. Practically all of the in-
vestigations which have been made of the influence of
nature and nurture on mental performance agree in attrib-
uting far more to original endowment than to environ-
ment. Common observation would itself suggest that the
social class to which the family belongs depends less on
chance than on the parents' native qualities of intellect and
character." (p. 115.)

lA. H. Arlitt. On the Need for Caution in Establishing Race Norms. Journal of

Applied Psychology, 1921, 5, pp. 179-183.

^L. M. Terman. The Measurement of Intelligence, Boston, 1916, Pp. 362.

AMERICAN INTELLIGENCE 189

One frequently hears the opinion expressed in scientific
circles that differences found between racial groups can
not be attributed to race unless the individuals examined
are drawn from the same social milieu. Miss Arlitt finds
native born white children of inferior and very inferior
social status above the Italian and negro children in in-
telligence, but attributes the larger differences found be-
tween the entire native white group and the Italian and
negro groups to the fact that three eighths of the native
white children come from homes of superior and very
superior social status. In the same way, Miss Murdoch
finds Jews living near East 110th St. and 2nd Ave. in New
York City not very inferior to native born whites living
in the mixed white and negro section around 8th Ave. and
140th St. The equalization of the environmental factor
is a necessary control in certain phases of scientific experi-
ments on race differences, but conclusions as to the intelli-
gence of racial groups must be drawn from samples taken
at random from the entire country. These conditions are
more nearly met by the army sampling of individuals in
the draft. Our samples of 81,465 native born individuals in
the white draft, of 12,492 foreign born individuals, and
23,596 negroes are drawn impartially from every section
of the country. If we selected our native born Americans
from those who live in the same squalid conditions in which
we find most of our negro and foreign bprn population, we
would not have a fair sample.

It is unfortunate that our army data classify foreign
born individuals only by country of origin, so that we have
no separate intelligence distributions for the Jews. Accord-
ing to the 1910 census, about 50% of the foreign born popu-
lation reporting Russia as their country of origin spoke
Hebrew or Yiddish, about 25% spoke Polish, less than 3%
spoke Russian, and the rest spoke Lithuanian, Lettish,

190 AMERICAN INTELLIGENCE

German, Finnish, Ruthenian and other tongues. From the
immigration statistics showing aliens admitted classified
according to race or people, we find about 10% (arriving
between 1900 and 1920) reported as Hebrew. It is fair to
assume that our army sample of immigrants from Russia
is at least one half Jewish, and that the sample we have
selected as Alpine i is from one fifth to one fourth Jewish.

Our figures, then, would rather tend to disprove the
popular behef that the Jew is highly intelligent. Immi-
grants examined in the army, who report their birthplace
as Russia, had an average intelligence below those from
all other countries except Poland and Italy. It is perhaps
significant to note, however, that the sample from Russia
has a higher standard deviation (2.83) than that of any
other immigrant group sampled, and that the Alpine
group has a higher standard deviation than the Nordic or
Mediterranean groups (2.60). If we assume that the Jewish
immigrants have a low average intelligence, but a higher
variability than other nativity groups, this would reconcile
our figures with popular belief, and, at the same time, with
the fact that investigators searching for talent in New York
City and California schools find a frequent occurrence of
talent among Jewish children. The able Jew is popularly
recognized not only because of his ability, but because he
is able and a Jew.

Our results showing the marked intellectual inferiority
of the negro are corroborated by practically all of the in-
vestigators who have used psychological tests on white and
negro groups. This inferiority holds even when a low in-
tellectual sampling of whites is made by selecting onfy

^There is no serious objection, from the anthropological standpoint, to classi-
fying the northern Jew as an Alpine, for he has the head form, stature, and color
of his Slavic neighbors. He is an Alpine Slav.

AMERICAN INTELLIGENCE 191

those who hve in the same environment, and who have had
the same educational opportunities. Professor Ferguson, ^
who has studied the problem most carefully, concludes that
in general 25% of the negroes exceed the median white. Our
figures show a greater difference than he estimates, less
than 12% of the negroes exceeding the average of the
native born white draft. Professor Ferguson also estimates
that 20% of pure negroes, 25% of negroes three quarters
pure, 30% of the true mulattoes, and 35% of the quad-
roons equal or exceed the average score of comparable
whites.

The discrepancies between data from various investiga-
tors as to the amount of difference between negroes and
whites probably result from different methods of selecting
whites. If we compare negroes only to those whites who
live in the same neighborhood, and who have had the
same educational opportunities, our differences are smaller
than those obtained by comparing samples of the entire
white and negro populations.

Some writers would account for the differences found
between white and negro by differences of educational
opportunity alone. The army tests showed the northern
negro superior to the southern negro, and this superiority
is attributed to the superior educational opportunities in
the North. The educational record of the negro sample we
are studying shows that more than half of the negroes from
the southern States did not go beyond the third grade, and
only 7% finished the eighth grade, while about half of the
northern negroes finished the fifth grade, and a quarter
finished the eighth grade. That the difference between the
northern and southern negro is not entirely due to school-

^G. O. Ferguson. The Mental Status of the American Negro. Scientific Monthly,
1921, 12, pp. 533-543.

192 AMERICAN INTELLIGENCE

ing, but partly to intelligence, is shown by the fact that
groups of southern and northern negroes of equal schooling
show striking differences in intelligence.

The superior intelligence measurements of the northern
negro are due to three factors : first, the greater amount of
educational opportunity, which does affect, to some ex-
tent, scores on our present intelligence tests; second, the
greater dmount of admixture of white blood; and, third,
the operation of economic and social forces, such as higher
wages, better living conditions, identical school privileges,
and a less complete social ostracism, tending to draw the
more inteUigent negro to the North. It is impossible to
dissect out of this complex of forces the relative weight of
each factor. No psychologist would maintain that the men-
tal tests he is now using do not measure educational oppor-
tunity to some extent. On the other hand, it is absurd to
attribute all differences found between northern and south-
ern negroes to superior educational opportunities in the
North, for differences are found between groups of the same
schooling, and differences are shown by beta as well as
by alpha.

At the present stage of development of psychological
tests, we can not measure the actual amount of difference
in intelligence due to race or nativity. We can only prove
that differences do exist, and we can interpret these differ-
ences in terms that have great social and economic signifi-
cance. The intellectual superiority of our Nordic group over
the Alpine, Mediterranean, and negro groups has been
demonstrated. If a person is unwilling to accept the race
hypothesis as developed here, he may go back to the
original nativity groups, and he can not deny the fact that
differences exist.

When our methods of measuring intellectual capacity
have been perfected, we will be in a position to determine

AMERICAN INTELLIGENCE 193

quantitatively the amount of race differences. Rough group
tests of the type we are now using will indicate the fact
that differences exist. However, while scientists are perfect-
ing their methods of examining, it would be well for them
to perfect their logic at the same time. Particularly mis-
leading and unsound is the theory that disregards all dif-
ferences found between racial groups unless the groups
have had the same educational and environmental oppor-
tunities.

This theory in its most extreme form is set forth by
Garth 1 as follows:

"The elements in a study of racial mental similarities or
differences must be these: (1) Two so-called races Rj and
R2, (2) an equal amount of educational opportunity, E,
which should include social pressure and racial patterns of
thought, and (3) psychological tests, D, within the grasp
of both racial groups. We should have as a result of our
experiment Ri E D equal to, greater than, or less than R^
E D. In this experiment the only unknown elements should
be Ri and R^. If E could be made equal the experiment
could be worked.

"This element of educational opportunity-nurture is the
one causing most of the trouble in racial psychology as an
uncontrollable element. It does not offer quite so much
difficulty in the study of sex differences, yet it is there
only in smaller degree than in racial differences, and as it
is controlled the *sex differences' tend to disappear. Since
this element of education, or nurture, cannot be eliminated
it would be safer to take for comparison such racial groups
as have had as nearly the same educational opportunity as
is possible having any disparity of this sort well in mind
when we interpret the results of the experiment. Having

^T. R. Garth. White, Indian and Negro Work Curves. Journal of Applied Psy -
chology, 1921, 5, pp. 14-25.

194 AMERICAN INTELLIGENCE

done this, we first take the complete distributions on the
scale of measurement for the groups as statements of the
true facts of the case, race for race. We then combine these
distributions into a total distribution of accompHshment
of all the races taken together to see if we have multimodal
effects. Should we find these effects we may conclude that
we have evidence of types, or racial types, and there should
in this case be one mode for each racial group. But should
the combined distribution for the several racial groups
reveal only one mode we may conclude that the test reveals
no types no real racial differences but rather similarities."
(p. 16.)

If intelligence counts for anything in the competition
among human beings, it is natural to expect that individ-
uals of superior intelligence will adjust themselves more
easily to their physical and social environment, and that
they will endow their children not only with material
goods, but with the ability to adjust themselves to the
same or a more complex environment. To select individuals
who have fallen behind in the struggle to adjust themselves
to the civilization their race has built as typical of that
race is an error, for their position itself shows that they
are, for the most part, individuals with an inferior heredi-
tary endowment.

In the same way, our educational institutions are them-
selves a part of our own race heritage. The average negro
child can not advance through an educational curriculum
adapted to the Anglo-Saxon child in step with that child.
To select children of equal education, age for age, in the
two groups, is to sample either superior negroes or inferior
whites.

The scientific problem is that of eliminating from the
tests used as measuring instruments those particular tests
which demonstrably measure nurture, and to measure.

AMERICAN INTELLIGENCE 195

with genuine tests of native intelligence, random or im-
partial samples from each race throughout the entire range
of its geographical and institutional distribution.

CONCLUSIONS

Our study of the army tests of foreign born individuals
has pointed at every step to the conclusion that the aver-
age intelligence of our immigrants is declining. This deteri-
oration in the intellectual level of immigrants has been
found to be due to two causes. The migrations of the Alpine
and Mediterranean races have increased to such an extent
in the last thirty or forty years that this blood now consti-
tutes 70% or 75% of the total immigration. The represen-
tatives of the Alpine and Mediterranean races in our
immigration are intellectually inferior to the representa-
tives of the Nordic race which formerly made up about
50% of our immigration. In addition, we find that we are
getting progressively lower and lower types from each
nativity group or race.

In the light of our findings in Sections IV and IX, it is pos-
sible to re-draw our curve (Figure 33) representing in-
crease of intelligence score with increasing years of residence
and to represent it truly as in Figure 46, which shows the
decline of intelligence with each succeeding period of im-
migration.

It is also possible to make a picture of the elements now
entering into American intelligence. At one extreme we
have the distribution of the Nordic race group. At the
other extreme we have the American negro. Between the
Nordic and the negro, but closer to the negro than to the
Nordic, we find the Alpine and Mediterranean types. These
distributions we have projected together in Figure 47.

Throughout this study all measurements have been made
in terms of averages and variability about the average. In
interpreting averages, we must never forget that they stand

197

WHITE OFFICERS I

WHITE DRAFT J^^
BORN IN ENGLAND

FOREIGN BORN
WHITE DRAFT

NATIVE BORN
"WHITE DRAFT

O CO

m Q

pq 12

en 03

NEGRO DRAFT I

OYEB^O

1897

16-20

-10

11-15
1898 1903

TO ijga

1902 1907

YEARS RESIDENCE IN UNITED STATES

TO
1912

0-5
1913

1917

Figure 46. The decline of intelligence with each succeeding period
of immigration. The apparent increase of intelligence with in-
creasing length of residence, as shown in Figure SS, has been
proved to be a progressive decrease in the intellectual level of
immigrants coming to this country in each succeeding five-year
period since 1902. The evidence indicates that the immigrants
prior to 1902 were intellectually equal to the native born white
draft. The army sample of "native born" includes, besides na-
tive born of native parentage, the native born of foreign or mixed
parentage. It is perhaps possible that the native born of native
parentage might have tested higher than 13.77. The position of
the white draft born in England is shown above. Although the
true position of the native born American may be a matter of
speculation, there is no doubt that the more recent immigrants
are intellectually closer to the negro than to the native born
white sample.

4J "^

ID 00

fir -M c3

53 n :t< ^

^.S?

^ p: c

p^ <

S 2 ^

83 O O

1^ 3 -^ p, PI

O ^-S ^ ^
g . P

<4-l c IS (D O "u

03 nd

0)0*33

Oh
O fl

faCrC

03
4J o o

rH ^ Oh

1^ IJ

202 AMERICAN INTELLIGENCE

for an entire distribution. Careless thinkers are prone to
select one or two striking examples of ability from a partic-
ular group, and then rest confidently in the belief that
they have overthrown an argument based on the total dis-
tribution of ability. The Fourth of July orator can con-
vincingly raise the popular belief in the intellectual level
of Poland by shouting the name of Kosciusko from a high
platform, but he can not alter the distribution of the intel-
ligence of the Polish immigrant. All countries send men of
exceptional ability to America, but the point is that some
send fewer than others.

Our distribution curve of intelligence includes ability as
well as defect. The English speaking Nordic group, for in-
stance, averages 13.84, and furnishes at one extreme about
40 men in 1000 who are above the average white officer,
while at the other extreme, the group furnishes about 8 in
1000 who are below an estimated "mental age" of eight. A
distribution further down the scale contributes more to the
lower orders of intelligence. The distribution of the intel-
ligence scores of the negro draft, for instance, indicates that
they contribute only 4 in 1000 above the average white
officer, while they give us 100 in 1000 below the approxi-
mate "mental age" of eight. The Alpine and Mediterranean
races give us only 5 or 10 in 1000 above the average ability
of the white officer, and about 40 in 1000 below the "mental
age" of eight. About 350 in 1000 of the Alpine and Mediter-
ranean types are below the average negro.

The intellectual characteristics of the immigration to the
United States as measured by the samples in the draft have
been reported in this study, first by country of birth, and
second by race. Parallel with the measurements of intel-
ligence, the figures on immigration have been presented.
To complete the picture, there is presented in Table 38
the population of the United States according to the 1920

Table No. 38
Population of the United States in 1920

Native White of Native Parentage. .58,421,957
Native White of Foreign Parentage . 15,694,539
Native White of Mixed Parentage . . 6,991 ,665

Total Native White 81,108,161

Foreign Born White 13,712,754

Negro 10,463,131

Indian : 244,437

Chinese 61,639

Japanese 111,010

All others :

Filipinos 5,603

Hindus 2,507

Koreans 1,224

Hawaiians 110

Malays 19

Siamese 17

Samoans 6

Maoris 2

Total all others 9,488

Total Population 105,710,620

NOTE: Clinton Stoddard Burr in America's Race Heritage (New York: The
National Historical Society, 1922, pp. 327) estimates that in 1920 there were
44,689,278 descendents of the old Colonial white stock.

203

204 AMERICAN INTELLIGENCE

census. 1 We have with us approximately 15 7/10 milHons
of individuals of foreign parentage, 7 millions of mixed
parentage, 13 7/10 millions of foreign born, and 10 Y2
millions of negroes. Roughly, in every 100 of our popula-
tion, ^^ are native born of native parentage, and the other
45 foreign born, or of foreign, mixed, or colored parentage.
The group of native born of native parentage includes
many children of the immigrants coming to this country
prior to 1890.

Our immigration figures show a very decided shift from
the Nordic in favor of the Alpine. The immigration between
1820 and 1890 probably never contained more than 50%
or 60% Nordic blood, and prior to 1820 there was very
little immigration. The earliest settlers were almost pure
Nordic types, and we may assume the existence by 1820
of a race as predominantly Nordic as that of England. This
recent change was, of course, reflected in the cross section
of the foreign born population taken at 1910, and which
constitutes the basis of our present immigration act re-
stricting immigration to 3% of the nationals then resident
here. A rough estimate of the racial composition of the
quotas from various countries admissable under the new
law shows about 35% Nordic blood, 45% Alpine blood and
20% Mediterranean blood in the annual stream of ap-
proximately 1/3 of a million that may enter.

There can be no doubt that recent history has shown a
movement of inferior peoples or inferior representatives of

iToo much reliance can not be'placed on the census returns for the foreign bom
white population. The 1910 census shows the foreign born white population as
13,345,545, while the 1920 census shows that population as 13,712,754, which
gives a net increase of 367,209. On the other hand, the figures of the Commis-
sioner General of Immigration show by actual count at the ports 5,725,811
aliens admitted and 2,146,994 aliens departed, leaving a net increase of 3,578,817
for the same period covered by the two censuses (1910 and 1920). Inasmuch as
the enumerators could not have missed three million, they are probably counted
among the native white.

AMERICAN INTELLIGENCE 205

peoples to this country. Few people realize the magnitude
of this movement or the speed with which it has taken
place. Since 1901, less than a single generation, it may be
estimated that about 10,000,000 Alpine and Mediterran-
ean types have come to this country. Allowing for the re-
turn of 1/3 or 3/8 of these, and using our army estimates
of intellectual ability, this would give us over 2,000,000
immigrants below the average negro.

We may consider that the population of the United
States is made up of four racial elements, the Nordic,
Alpine, and Mediterranean races of Europe, and the negro.
If these four types blend in the future into one general
American type, then it is a foregone conclusion that this
future blended American will be less intelligent than the
present native born American, for the general results of the
admixture of higher and lower orders of intelligence must
inevitably be a mean between the two.

If we turn to the history of races, we find that as a general
rule where two races have been in contact they have inter-
mingled, and a cross between the two has resulted. Europe
shows many examples of areas where the anthropological
characteristics of one race shade over into those of another
race where the two have intermixed, and, indeed, in coun-
tries such as France and Switzerland it is only in areas that
are geographically or economically isolated that one finds
types that are relatively pure. The Mongol-Tatar element
in Russia is an integral part of the population. The Mediter-
ranean race throughout the area of its contact with the
negro has crossed with him. Some of the Berbers in
Northern Africa show negroid characteristics, and in India
the Mediterranean race has crossed with the Dravidians
and Pre-Dra vidian negroids. The population of Sardinia
shows a number of negroid characteristics. Turn where we
may, history gives us no great exception to the general rule

206 AMERICAN INTELLIGENCE

that propinquity leads to opportunity and opportunity to
intermixture.

In considering racial crosses, Professor Conklini states
that "It is highly probable that while some of these hy-
brids may show all the bad qualities of both parents, others
may show the good qualities of both and indeed in this re-
spect resemble the children in any pure-bred family. But
it is practically certain that the general or average results
of the crossing of a superior and an inferior race are to
strike a balance somewhere between the two. This is no
contradiction of the principles of Mendelian inheritance
but rather the application of these principles to a general
population. The general effect of the hybridization of races
can not fail to lead to a lowering of the qualities of the
higher race and a raising of the qualities of the lower one."
(pp. 50-51.)

And as to the possibility of a cross between races in the
future. Professor Conklin writes: "Even if we are horrified
by the thought, we cannot hide the fact that all present
signs point to an intimate commingling of all existing hu-
man types within the next five or ten thousand years at
most. Unless we can re-establish geographical isolation
of races, we cannot prevent their interbreeding. By rigid
laws excluding immigrants of other races, such as they have
in New Zealand and Australia, it may be possible for a time
to maintain the purity of the white race in certain coun-
tries, but with constantly increasing intercommunications
between all lands and peoples such artificial barriers will
probably prove as ineffectual in the long run as the Great
Wall of China. The races of the world are not drawing apart
but together, and it needs only the vision that will look

1 Edwin G. Conklin. The Direction of Human Evolution. New York, 1921, pp. 247.

AMERICAN INTELLIGENCE 207

ahead a few thousand years to see the blending of all racial
currents into a common stream." (p. 52.)

If we frankly recognize the fact that the crossing of races
in juxtaposition has always occurred in the past, what
evidence have we that such crosses have had untoward con-
sequences? Our own data from the army tests indicate
clearly the intellectual superiority of the Nordic race
group. This superiority is confirmed by observation of this
race in history. The Alpine race, according to our figures,
which are supported by historical evidence, seems to be
considerably below the Nordic type intellectually. How-
ever, our recruits from Germany, which represents a Nordic-
Alpine cross, are about the same as those from Holland,
Scotland, the United States, Denmark, and Canada, coun-
tries which have on the whole a greater proportion of Nordic
blood than Germany. Again, the Nordic and Alpine mix-
ture in Switzerland has given a stable people, who have
evolved, in spite of linguistic differences, a very advanced
form of government. The evidence indicates that the
Nordic-Alpine cross, which occurred in Western Europe
when the Nordics overwhelmed the Alpines to such an ex-
tent that the type was completely submerged and not
re-discovered until recently, has not given unfortunate re-
sults.

This evidence, however, can not be carried over to indi-
cate that a cross between the Nordic and the Alpine Slav
would be desirable. The Alpines that our data sample come
for the most part from an area peopled largely by a
branch of the Alpine race which appeared late and radiated
from the Carpathian Mountains. It is probably a different
branch of the Alpine race from that which forms the prim-
itive substratum of the present population of Western
Europe. Our data on the Alpine Slav show that he is in-
tellectually inferior to the Nordic, and every indication

208 AMERICAN INTELLIGENCE

would point to a lowering of the average intelligence of the
Nordic if crossed with the Alpine Slav. There can be no
objection to the intermixture of races of equal ability, pro-
vided the mingling proceeds equally from all sections of
the distribution of ability. Our data, however, indicate that
the Alpine Slav we have imported and to whom we give
preference in our present immigration law is intellectually
inferior to the Nordic type.

The Mediterranean race at its northern extension blends
with the Alpine very considerably, and to a less extent with
the Nordic. At the point of its furthermost western expan-
sion in Europe it has crossed with the primitive types in
Ireland. Throughout the area of its southern and eastern
expansion it has crossed with negroid types. In this con-
tinent, the Mediterranean has crossed with the Amerind
and the imported negro very extensively. In general, the
Mediterranean race has crossed with primitive race types
more completely and promiscuously than either the Alpine
or the Nordic, and with most unfortunate results.

We must now frankly admit the undesirable results which
would ensue from a cross between the Nordic in this coun-
try with the Alpine Slav, with the degenerated hybrid
Mediterranean, or with the negro, or from the promiscu-
ous intermingling of all four types. Granted the undesirable
results of such an intermingling, is there any evidence
showing that such a process is going on.^^ Unfortunately the
evidence is undeniable. The 1920 census shows that we
have 7,000,000 native born whites of mixed parentage, a
fact which indicates clearly the number of crosses between
the native born stock and the European importations.

The evidence in regard to the white and negro cross is
also indisputable. If we examine the figures showing the
proportion of mulattoes to a thousand blacks for each
twenty year period from 1850 to 1910, we find that in 1850

AMERICAN INTELLIGENCE 209

there were 126 mulattoes to a thousand blacks, 136 in 1870,
179 in 1890 and 264 in 1910. This intermixture of white and
negro has been a natural result of the emancipation of the
negro and the breaking down of social barriers against him,
mostly in the North and West. In 1850, the free colored
population showed 581 mulattoes to a thousand blacks as
against 83 in the slave population. At each of the four cen-
suses (1850, 1870, 1890 and 1910) the South, where the
social barriers are more rigid than elsewhere, has returned
the smallest proportion of mulattoes to a thousand blacks.
The 1910 census showed 201 in the South, 266 in the
North and 321 in the West, and the West has returned the
highest proportion at each of the censuses except 1850.

We must face a possibility of racial admixture here that
is infinitely worse than that faced by any European country
to-day, for we are incorporating the negro into our racial
stock, while all of Europe is comparatively free from this
taint. It is true that the rate of increase of the negro in this
country by ten year periods since 1800 has decreased rather
steadily from about 30% to about 11%, but this declining
rate has given a gross population increase from approximate-
ly 1,000,000 to approximately 10,000,000. It is also true that
the negro now constitutes only about 10% of the total
population, where he formerly constituted 18% or 19%
(1790 to 1830), but part of this decrease in percentage of
the total population is due to the great influx of immi-
grants, and we favor in our immigration law those coun-
tries 35% of whose representatives here are below the aver-
age negro. The declining rate of increase in the negro
population from 1800 to 1910 would indicate a correspond-
ingly lower rate to be expected in the future. From 1900 to
1920 the negro population increased 18.4%, while the na-
tive born white of native parents increased 42.6%, and the
native born white of foreign parents increased 47.6%. It is

210 AMERICAN INTELLIGENCE

impossible to predict at the present time that the rate of
infiltration of white blood into the negro will be checked
by the declining rate of increase in the negro blood itself.
The essential point is that there are 10,000,000 negroes here
now and that the proportion of mulattoes to a thousand
blacks has increased with alarming rapidity since 1850.

According to all evidence available, then, American in-
telligence is- declining, and will proceed with an accelerat-
ing rate as the racial admixture becomes more and more
extensive. The decline of American intelligence will be more
rapid than the decline of the intelligence of European
national groups, owing to the presence here of the negro.
These are the plain, if somewhat ugly, facts that our study
shows. The deterioration of American intelligence is not
inevitable, however, if public action can be aroused to pre-
vent it. There is no reason why legal steps should not be
taken which would insure a continuously progressive up-
ward evolution.

The steps that should be taken to preserve or increase
our present intellectual capacity must of course be dictated
by science and not by political expediency. Immigration
should not only be restrictive but highly selective. And the
revision of the immigration and naturalization laws will
only afford a slight relief from our present difficulty. The
really important steps are those looking toward the pre-
vention of the continued propagation of defective strains
in the present population. If all immigration were stopped
now, the decline of American intelligence would still be
inevitable. This is the problem which must be met, and our
manner of meeting it will determine the future course of
our national life.

Uoiversity of Torooto
library

DO NOT

REMOVE

THE

CARD

FROM

THIS

POCKET

Acme Library Card Pocket

Uader Pat. "Ref. Index FUt"

Made by LIBRARY BUREAU

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "A study of American intelligence"

See other formats